Intraspecific variation in tree growth responses to neighbourhood composition and seasonal drought in a tropical forest

Abstract

Functional traits are expected to provide insights into the abiotic and biotic drivers of plant demography. However, successfully linking traits to plant demographic performance likely requires the consideration of important contextual and individual‐level information that is often ignored in trait‐based ecology.
Here, we modelled 8 years of growth from 1,138 individual trees in 36 tropical rain forest species. We compared models of tree growth parameterized using individual‐level versus species mean trait data. We also compared models that considered regional climatic, local biotic and whole‐plant allocation contexts to those that do not.
Our analyses show that growth models parameterized using individual‐level trait information outperformed those that used species mean trait information and that these models often contradicted one another indicating that the common practice of using species mean trait data requires more scrutiny. Additionally, we found that models including climatic, biotic and allocation contexts outperformed those that did not and provided nuanced insights into the drivers of tree growth in a tropical forest.
Synthesis. Here we have shown that the development of models of tree demographic performance upon the basis of traits can be improved through a consideration of individual‐level trait variation as well as phenotypic and climatic contexts. We highlight that our ability to understand the drivers of tree population and community structure and dynamics in current and in future climates will be limited if contextual and individual‐level data remains understudied.

Full text

26
|
wileyonlinelibrary.com/journal/jec Journal of Ecology. 2021;109:26–37.© 2020 British Ecological Society
Received: 26 February 2020
|
Accepted: 11 May 2020
DOI : 10.1111/136 5-2745.134 39
RESEARCH ARTICLE
Intraspecific variation in tree growth responses to
neighbourhood composition and seasonal drought in a tropical
forest
Jie Yang1,2 | Xiaoyang Song1,3 | Jenny Zambrano4 | Yuxin Chen5 | Min Cao1 |
Xiaobao Deng1 | Wenfu Zhang1 | Xiaofei Yang1 | Guocheng Zhang1 | Yong Tang1 |
Nathan G. Swenson6
1CAS Key Laborator y of Tropical Forest Ecology, Xishuangbanna Tropical Botanical Garden, Chinese Academy of Sciences , Mengla , China; 2Center of Plant Ecol ogy,
Core Botanical Gardens , Chinese Academy of Sciences, Mengla, China; 3Center of Conse rvation Biology, Core Botanical Gardens, Chinese Academy of Science s,
Mengla, China; 4School of Biological Sciences, Washington State Uni versit y, Pullman, WA, USA; 5Key Laboratory of the Coastal an d Wetland Ecos ystems (Ministry of
Education), College of the Environment & Ecology, Xiamen University, Xiamen, China and 6Depar tment of Biology, Universit y of Maryland, College Park, MD, USA
Jie Yang and X iaoyang Song con tributed equ ally to th is work.
Correspondence
Min Cao
Email: ca om@xtbg.ac.cn
Funding information
Nationa l Natural Science Foundation of
China, G rant/Award Number: 31800353
and 31670442; Strategic Priority Res earch
Program of the Chinese Academy
of Sciences, Grant/Award Number:
XDB3100 00 00; N SF US-C hina Dimensions
of Biodiversit y grant s, Grant/Award
Number : DEB-1241136 and DEB-1046113;
The CAS 135 program, Grant/Award
Number : 2017XTBG-T01; The Chinese
Academy of Sciences Youth Innovation
Promotion Association, Grant/Award
Number : 2016352; The S outhe ast Asia
Biodiversity Research Institute, Chinese
Academy of Sciences, Grant/Award
Number : Y4ZK111B01; The West Light
Foundation of the Chinese Academy of
Science s; Ten Thousand Talents Program of
Yunnan, Grant/Award N umber : YNWR-
QNBJ-2018-30 9; Xishua ngbanna Station for
Tropical Rain Forest Ecosystem Studies
Handling Editor: Andrew Hector
Abstract
1. Functional traits are expected to provide insights into the abiotic and biotic driv-
ers of plant demography. However, successfully linking traits to plant demo-
graphic performance likely requires the consideration of important contextual and
individual-level information that is often ignored in trait-based ecology.
2. Here, we modelled 8 years of growth from 1,138 individual trees in 36 tropical rain
forest species. We compared models of tree growth parameterized using individual-
level versus species mean trait data. We also compared models that considered re-
gional climatic, local biotic and whole-plant allocation contexts to those that do not.
3. Our analyses show that growth models parameterized using individual-level trait
information outperformed those that used species mean trait information and
that these models often contradicted one another indicating that the common
practice of using species mean trait data requires more scrutiny. Additionally, we
found that models including climatic, biotic and allocation contexts outperformed
those that did not and provided nuanced insights into the drivers of tree growth in
a tropical forest.
4. Synthesis. Here we have shown that the development of models of tree demo-
graphic performance upon the basis of traits can be improved through a consider-
ation of individual-level trait variation as well as phenotypic and climatic contexts.
We highlight that our ability to understand the drivers of tree population and com-
munity structure and dynamics in current and in future climates will be limited if
contextual and individual-level data remains understudied.
KEYWORDS
drought, functional traits, integrative traits, neighbourhood interactions

|
27
Journal of Ecolog
y
YANG et Al.
1 | INTRODUCTION
Climate is a fundamental regional-scale driver of plant demographic
performance (Chmura et al., 2011; Pretzsch, Biber, Schütze, Uhl, &
Rötzer, 2014). Accordingly, rapid changes in climate are predicted to
drive notable shifts in the composition and dynamics of vegetation
(Kelly & Goulden, 2008; McKenney, Pedlar, Lawrence, Campbell, &
Hutchinson, 2007). Understanding the factors that drive the de-
mographic performance of trees will, therefore, help to elucidate
forest community organization and dynamics under future climatic
conditions. On local scales, biotic interactions, often measured as
the degree of crowding by neighbours, are also a key determinant of
tree demographic performance (Canham, Lepage, & Coates, 200 4;
Uriarte et al., 2010). While both regional-scale climate and local-scale
biotic interactions are appreciated in the literature (Nowacki &
Abrams, 2015; Pederson et al., 2015), their interactive effects on
tree demography have received surprisingly little attention despite
their importance for forest dynamics (Clark, Bell, Kwit, & Zhu, 2014;
Ford et al., 2017; Newber y & Stoll, 2013; O'Brien, Reynolds, Ong, &
Hector, 2017; Uriarte, Muscarella, & Zimmerman, 2018).
Plant functional traits, such as wood density, leaf mass per area
(LMA) and seed mass, can serve as valuable indicators of how climatic
and biotic interactions influence tree demography and, ultimately,
forest dynamics (Swenson, 2013; Swenson et al., 2012). Specifically,
functional traits should determine the demographic performance of
an individual in a given abiotic and biotic environment. This differen-
tial demographic performance should scale up to determine popula-
tion and community structure and dynamics (Arnold, 1983; McGill,
Enquist, Weiher, & Westoby, 2006; Poorter et al., 2008). Although
trait–demography relationships provide evidence linking organismal
traits to the abiotic and biotic environment, in many cases, widely
measured plant functional trait s such as LMA are not correlated or
are weakly correlated with tree demographic performance (Hérault
et al., 2011; Paine et al., 2015; Yang, Cao, & Swenson, 2018).
Recently, Yang et al. (2018) argued that there are three core rea-
sons underlying these weak relationships—a lack of environmental
context, failure to integrate trait data and a failure to consider in-
traspecific variation. We will briefly consider each of these reasons
here. First, the abiotic and biotic environment in which an individual
tree is located must be simultaneously considered. That is, the re-
gional scale climate may interact with the local neighbourhood com-
position to drive tree demographic per formance such that the role
of each of these contexts cannot be understood without the other.
A recent example of this was produced by Zambrano, Marchand,
and Swenson (2017) who showed that the strength of intraspecific
negative density dependence interacts with the regional scale cli-
matic suitability in which an individual tree exists. While this work
is important, climate varies substantially within and between years
and a stronger consideration of this climatic dynamism is needed in
neighbourhood studies of tree demography ( Yang et al., 2018).
A second pot ent ia l reaso n fo r we ak tr ai t–p erf orm ance rela tio ns hip s
in tre e eco l ogy is tha t le af-l eve l tra i t va lue s are no t integ r ate d to prov ide
an integrated phenotype that can be used for predicting performance
(Yang et al., 2018). For example, easily measured leaf traits (e.g. LMA)
believed to be linked to resource acquisition strategies (Reich, Walters,
& Ellswor th, 1997; Wright et al ., 200 4) are often wea kl y correlated with
tree growth rates (e.g. Lasky, Uriarte, Boukili, & Chazdon, 2014; Paine
et al., 2015; Poorter et al., 2008; Wright et al., 2010). This could be
because a proxy of resource capture rate per unit leaf area or mass
will fail to predict grow th rates without knowing the relative allocation
to leaf area or mass for the whole plant. While the need to integrate
traits and allocation is understood by some (e.g. Enquist et al., 2007;
Poorter, 1989; Poorter, Niinemets, Poorter, Wright, & Villar, 2009), it
is often ignored in trait-based studies of tree demography. One poten-
tially powerful, and feasible, way to perform such an integration would
be to place LMA into a whole crown allocation context (Yang et al.,
2018) . Th e LM A of a plant is positi ve ly rela te d to lea f life spa n an d ne ga-
tively related to resource capture rates (Moles, 2018; Reich et al., 1999;
Wright et al., 2004; but see Osnas, Lichstein, Reich, & Pacala, 2013).
However, LMA represents the strategy of a single leaf without context
regarding whole crown all ocation . Th us , th e pr od uct of LMA and crown
volume (LMA × crown volume = LMAadjusted) would provide a rough
estimate of the total allocation to canopy leaf mass. Previous work has
shown a strong relationship between total photosynthetic mass and
plant growth rate (Niklas & Enquist, 2001). This indicates that a trait
like LMA , which often fails to predict tree per formance, would become
a stronger predictor if adjusted for crown volume.
A final reason why easily measured functional traits often fail to
predict tree per formance is that species mean trait values and mean
demographic per formance are analysed. That is intraspecific varia-
tion in traits and performance is often not considered. This is despite
the considerable amount of intraspecific variation in plant traits and
demo gra phic pe r fo rma nce in natu ral for est syst ems. Th at said , intra -
specific variation has is becoming increasingly utilized in trait-based
ecology. For example, there is evidence that our understanding of
communit y assembly at local scales depends critically and relies on
our understanding of intraspecific trait variation (e.g. Jung, Violle,
Mondy, Hof fmann, & Muller, 2010; Violle et al., 2012). An omission
of intraspecific variation likely results in an obscured relationship be-
tween traits and demography (e.g. Iida et al., 2014; Liu et al., 2016;
von Oheimb et al., 2011; Yang et al., 2018).
In sum, the demographic performance of plants arises from rela-
tionships with the abiotic and biotic environment and their interaction.
These interactions should be related to traits, but these relationships
are of ten weak. These weak relationships may be due to a failure to
simultaneously consider the dynamic climatic context and biotic con-
text, a failure to integrate single traits into a whole plant context and
an over-reliance on species average trait values (Yang et al., 2018).
Over 40% of tropical forests world-wide are subject to seasonal
water stress (Miles et al., 2006; Myneni et al., 2007). Plants in these
forests generally downregulate photosynthesis or shed leaves in the
dry season (Restrepo-Coupe et al., 2013). These alternative photo-
synthetic and phenological strategies have different implications for
individual growth (Enquist & Leffler, 2001; Grogan & Schulze, 2012).
The investigation of the combined effects of climate and local com-
petition on tree demography and their relative contributions to forest

28
|
Journal of Ecology
YANG et Al.
dy nam ic s has be en usef ul (e. g . Can ha m , Pa pai k, Ur ia r te, & Mc Will iam s,
2006; Clark et al., 2014; Zambrano et al., 2017). However, these previ-
ous studies have typically examined tree performance during a single
ti m e-p e rio d wi t hou t con s ide rat ion of rep e ate d tem por al ch a nge s in th e
abiotic environment. However, seasonal changes in temperature and
precipitation during wet or dry seasons strongly impact plant phenol-
ogy, plant growth and water relations (Zeppel, Wilks, & Lewis, 2014).
Therefore, it is likely that average annual growth rates aggregated from
data taken over long periods and correlated with summar y statis tic s of
annual climate mask substantial shorter-term variation in tree grow th
rates in response to intra-annual climate variability.
Here we combine detailed individual-level tree growth, biotic
neighbourhood and trait data with local climatic data over a period
of 9 years in a seasonal tropical rain forest. The specific questions
we asked in the study are: (a) Do models of individual tree growth
with individual-level trait data outper form those with species aver-
age trait values? (b) Does integrating leaf trait data (i.e. LMA) into a
crown contex t (i.e. LM Aadjusted) provide better models of tree growth
than those parameterized using non-integrative traits? and (c) What
are the interactive effects between traits, neighbourhood and cli-
matic context on tree growth?
2 | MATERIALS AND METHODS
2.1 | Study site
The study was conducted in the 20-ha Xishuangbanna forest dy-
namics plot (FDP) in a seasonal tropical rainforest of Southwest
China (21°37′08″N, 101°35′07″E; Figure S1). This forest is domi-
nated by large individuals of Parashorea chinensis (Dipterocarpaceae).
Monsoonal effects are a dominant driver of rainfall and air tempera-
ture on the local climate and produce marked seasonal patterns, with
distinct alternations between the dry season (November–April) and
the wet season (May–October). The dry season is characterized by
approximately five continuous months with lit tle precipitation and
lower temperature (Figure S2). The annual rainfall averages 1,493 mm
in the fo re st, of which 1, 256 mm (84%) occu rs in the we t se as on (Ca o,
Zou, War re n, & Zhu, 20 06). The st ron gl y sea sonal precip ita ti on dr ives
more ex te nsive varia tion of tre e gr ow th wit hin year than acr oss years
(VA Rseason = 0.1, VARyear = 0.07; p < 0.01). The Xishuangbanna FDP
was established in 20 07, and censuses are carried out every 5 years.
All freestanding woody ste ms ≥1 cm diam eter at breast height (DBH),
that is, 130 cm from the ground, are measured, tagged , ide ntified and
mapped. There were a total of 468 species and 95,946 individuals in
the plot first census. A detailed description of the climate, geology
and flora of this plot can be found in Cao et al. (2008).
2.2 | Growth rate
This study uses a subset of trees in the forest plot that have
growth monitored on a fine temporal scale using dendrometer
bands. Dendrometers offer the potential to provide continuous
high-resolution measurements of tree radius changes. Specifically,
stainless steel dendrometer bands were installed in the plot from
January 2009 following a standardized protocol (Muller-Landau &
Dong, 2008). Digital calipers were used to record the window size on
the dendrometer for each measurement. The first re-measurement
was conducted in August 2009 and, since then, the dendrometers
have been re-measured every 3 months. To guarantee data accu-
racy, we discarded the cases where (a) the tree had damaged den-
drometer bands; (b) the tree died during the study period; (c) the tree
was within 15 m from the plot edge; or (d) the tree had incomplete
trait or growth data. Furthermore, in order to compare models that
included intraspecific or interspecific trait data, we used those spe-
cies with an abundance of ≥15 in the forest plot. For each focal tree,
tree circumference increments were transformed into diameter first,
we then calculated its annual absolute diameter growth (AGR) dur-
ing the dry season and wet season per year from November 2009
to May 2018. All AGR values that exceed four standard deviations
from the mean were discarded (c. 5%), we finally obtained a dataset
containing 65,506 growth records for 1,138 individual trees belong-
ing to 36 species (Table S1; Figure S3). Here we refer to the growth
rate within each season as ‘seasonal growth’, while we refer to the
growth rate during a year as ‘annual growth’.
2.3 | Functional traits
We measured four functional traits for each tree with a dendrom-
eter band using the standard protocols described in Cornelissen
et al. (2003) with a few exceptions noted below (Table S2). Here we
included tree height, wood-specific resistance (WSR), crown adjusted
le af mass pe r are a (LM Aadjusted) and leaf area, which are thought to be
strong indicators of plant functional strategies and are expected to
be linked to individual tree growth in response to seasonal drought
(Westoby & Wright, 2006; Wright et al., 2010). Specifically, we
represented total allocation to the crown as LMA × crown volume,
called LMAadjusted in this study. The dimensions of tree crowns were
calculated to estimate canopy volume. Tree height (H), height of the
lowest foliage (Hf), two orthogonal widths of crown (W1 and W2)
were measured using a laser telemeter attached to an altimeter pole
following the protocols described in Iida et al. (2012) to calculate the
volume of an ellipsoid:
where W1 is the crown width, W2 is the crown length, H is the tree
height and Hf is the height of the lowest foliage.
Th e WSR wa s me asu red fo r each ind i v idua l usi n g a Resi s t ogra ph
(Rinntech Co.), an electronically controlled drill that measures the
relationship between drilling resistance and stem density (Isik &
Li, 2003; Yang et al., 2014). The WSR of an individual is strongly
correlated with the more commonly measured trait of wood
(1)
Canopy volume
=
4
3
𝜋
1
2
W1
1
2
W2
1
2
(H−Hf)
,

|
29
Journal of Ecolog
y
YANG et Al.
density, while being a less destructive measurement on individuals
undergoing long-term monitoring (Isik & Li, 2003). A total of 3–15
leaves, that did not have any obvious symptoms of pathogen or her-
bivore attack and without substantial cover of epiphylls, were ran-
domly collected at the tree canopy for each individual (Cornelissen
et al., 2003). Leaf area was measured using scanner and ImageJ
software by r package Lea farea (Katabuchi, 2015). The LMA was
measured by the dry leaf mass divided by leaf area. The measure-
ment of leaf functional traits followed the same methodologies
as used by Yang et al. (2014). Pearson correlation coefficients for
pairwise correlation between the functional trait data was show in
the Table S3. We used each trait in both the individual- and spe-
cies-level models. All trait data were centring transformed for nor-
mality prior to analysis.
2.4 | Local climatic data
Water availability and temperature are major climatic factors
in monsoon systems that drive seasonal drought (Figure S1;
Table S4). Vapour pressure deficit ( VPD) summarizes the joint eco-
logical effects of precipitation and temperature (McDowell et al.,
2008; Will, Wilson, Zou, & Hennessey, 2013). The VPD has been
shown to accurately capture drought effects on tree performance
in the previous work (Eamus, Boulain, Cleverly, & Breshears, 2013;
McDowell et al., 2008; Uriarte, Lasky, Boukili, & Chazdon, 2016),
with higher VPD potentially exacerbating the physiological stress
on plants during drought events by increasing plant water loss
and/or reducing carbon uptake. The monthly VPD data were avail-
able from the automated meteorological data acquisition system
of National Field Scientific Observation and Research Station in
the Xishuangbanna Forest Ecosystem. Models for wet and dry sea-
son growth in our work used VPD data from 3-month windows
as follows: December–February and March–May for dry season
growth and June–August and September–November for wet sea-
son growth (Figure 1).
2.5 | Neighbourhood variables
To assess the local neighbourhood effects on focal tree growth,
neighbourhood variables were measured. We calculated a neigh-
bourhood crowding index (NCI) for each focal tree (i) based on the
size and spatial distance of its neighbours within a fixed radius (15 m)
as follows:
where dij is the spatial distance between a focal individual i and a
neighbour tree j and DBHj is the DBH of neighbour tree j. The distance
of 15 m was used as it is approximately the spatial scale at which bi-
otic interactions can no longer be detected in this forest plot (Yang
et al., 2014).
(2)
NCI
i=
j
∑
j=1,i≠j
DBH
2
j
d2
ij
,
FIGURE 1 Three monthly variation in average vapour pressure deficit and 36 focal species growth rate in Xishuangbanna seasonality
forest from 2010 to 2018
0.0
0.5
1.0
1.5
May
10
Nov
10
May
11
Nov
11
May
12
Nov
12
May
13
Nov
13
May
14
Nov
14
May
15
Nov
15
May
16
Nov
16
May
17
Time
Speciesgrowthrate
Vapour pressure deficit
Values

30
|
Journal of Ecology
YANG et Al.
To better understand the relationship between tree growth and
neighbourhood competition, we calculated separate indices to quan-
tify the effect of neighbourhood hierarchical trait distance (NCIH) and
neighbourhood absolute trait distance (NCIS). NCIH and NCIS are as-
sociated with trait-mediated ranks in competitive abilities and niche
differentiation, respectively (Kunstler et al., 2012; Lasky et al., 2014):
and
where Fi and Fj are the values of the functional trait of interest for the spe-
cies of the focal individual (i) and neighbours (j). The focal individual trait
used the data of each tree from which growth was measured, while the
neighbours’ trait used the average trait value calculated for each species.
2.6 | Statistical analyses
2.6.1 | Effects of intraspecific and interspecific trait
variation on annual tree growth
One goal of this study was to evaluate whether individual-level trait
values were a better predictor of tree grow th than species mean
trait values (i.e. question a). To address this question, we constructed
two sets of growth models, one using individual-level trait data and
the other with species-level mean trait values as predictors. The full
model had the following form (Table 1):
where Yi,t represent s the expected growt h rate of individual i in year
t and β0 is the intercept for all individuals. The latter four terms in
Equation (5) include the functional traits of the focal tree (i.e. actual
individual trait values for the intraspecific model or a mean for the
(3)
NCIH
i=
j
∑
j
=1,
i
≠
j
(Fi−Fj)DBH
2
j
d2
ij
,
(4)
NCIS
i=
j
∑
j=1,i≠j|
|
|
Fi−Fj|
|
|
DBH
2
j
d2
ij
,
(5)
Y
i,t=
𝛽
0+
𝛽
1
trait
i+
𝛽
2
trait
i×
NCI
i+
𝛽
3
traiti
×NCIH
i
+
𝛽4traiti
×NCIS
i
+
𝜄t
+
𝜑s
+
𝛾i,
Individual traits model Species mean trait model
Model name Best model Model name Full model
Fixed effects b [95% CI] Fixed effects b [95% CI]
Intercept 0.31 [−0.15, 0.77] Intercept 3.62 [3.3, 3.94]
LMAadjusted 2.83 [2.21, 3.45] WSR −7. 7 3 [−9.1, −6.32]
NCI −1. 2 5 [−1.55, −0.95] NCI −4.84 [−5.3, −4.39]
NCIH.LMAadjusted 0.72 [−0.08, 1. 52] NCIH.wsr −0.18 [−0.6, 0.19]
NCIS. LMAadjusted 0.35 [−0.17, 0.87] NCIS.wsr 1.05 [0.81, 1.29]
LMAadjusted:NCI −4. 51 [−5.2, −3.82] WSR:NCI 10. 84 [8.9, 12.8]
LMAadjusted:
NCIH.LMAadjusted
−1.47 [−2.35, −0.59] WSR:NCIH.wsr 0.82 [−1.1, 2.78]
LMAadjusted:
NCI.LMAadjusted
−0.53 [−1.47, 0.41] WSR:NCI.wsr −4.28 [−5.3, −3.24]
Random effects σ2Random effects σ2
Yea r 0.06 Year 0.05
Species 0.01 Species 0.02
Plot 0.01 Plot 0.01
Residual 2.94 Residual 3.15
Fixed factors 0.08 Fixed factors 0.13
R2
GLMM(m) (%) 10.3 R2
GLMM(m) (%) 3.8
R2
GLMM(c) (%) 12 R2
GLMM(c) (%) 5.9
AIC 33,583 AIC 34,191
BIC 33,6 61 BIC 34, 268
Note: The LMAadjusted represent s the LMA × crown volume, the WSR represents wood-specific
resistance. Neighbourhood variables included neighbourhood competition index (NCI), trait
hierarchies (NCIH) and absolute tr ait differences with neighbours (NCIS). The CI is confidence
interval, σ2 is the variance components; The R2
GLMM(m) is marginal R2 which is the variance
explained by the fixed fac tors; the R2
GLMM(c) is conditional R2 which is the variance explained by the
entire model. AIC is Akaike information criterion value; BIC is Bayesian information criterion. The
AIC and BIC values were calculated using ML and REML estimations. The bold values represent
statis tically significant results.
TABLE 1 Comparison of the best
mixed-effects models, for the ef fects
of functional traits, neighbourhood and
climate manipulations on tree growth,
using individual- or species-level trait data

|
31
Journal of Ecolog
y
YANG et Al.
species for the interspecific model), with the covariates correspond-
ing to the ef fect s of NCI, NCIH and NCIS, respectively (Lasky et al.,
2014). We utilized normally distributed random effects for years,
species and 20 × 20 m subplot s as
𝜄t
,
𝜑s
and
𝛾i
respectively. The NCI,
NCIH and NCIS var iab les were all st andardized to a mean of zero and
one standard deviation to facilitate interpretation and comparisons
among parameters.
We utilized 20 × 20 m sub pl ots as ran dom effects in the mode l to
account for any potential autocorrelation in growth between trees
in the same subplot. We did not use a larger spatial scale for this
random effect because very little-to-no correlation in growth could
be found beyond 10 m in distance.
To compare models using individual-level and species-level mean
functional traits, we used the variance explained (R2) of the GLMMs
to show the absolute value for the goodness-of-fits of each model
(Nakagawa & Schielzeth, 2013). To guarantee the validity of these
model comparisons, we used same combinations of fixed effects for
the models being compared and we used the same number of trees
(i.e. the same sample size) in each model. Two kinds of R2
GLMM for
the variance components were calculated (Table 1): the marginal R2
(R2
GLMM(m)) which is the variance explained by fixed factors and the
conditional R2 (R2
GLMM(c )) which is the variance explained by the en-
tire model (Nakagawa & Schielzeth, 2013). We also calculated Akaike
Information Criterion (AIC) and Bayesian Information Criterion (BIC)
values using maximum likelihood (ML) and restricted maximum like-
lihood (REML) estimations.
2.6.2 | Effects of climate on seasonal tree growth
To assess the effects of trait–climate–neighbourhood interactions
on tre e growth (i.e. question c), we generated seasonal growth mod-
els. We included average dry and wet VPD for each growing sea-
son as one of the independent variables, which allowed us to assess
potential effects of climate on tree seasonal growth variation. We
adde d co mb inations of dry- and wet- sea so n as cov ariates in th e sea-
sonal growth models. Our grow th models included the interaction
between VPD and the local neighbourhood crowding variable (NCI)
as competition may intensify when resources (e.g. water availabil-
ity) become more limited. The combined effects of climate and local
neighbourhood were described as follows:
where Zi,t and VPDt are the growth rate and VPD in season t respec-
tively. The linear predictor Zi,t included main climate and neighbour-
hood crowding, as well as their interactions. Normally distributed
random effects for species and 20 × 20 subplots are
𝜏t
and
𝜑s
respec-
tively. As noted above, we utilized a subplot random effect to account
for the very little spatial autocorrelation that exists bet ween immedi-
ately neighbouring trees.
To assess whether an integrative trait (LMAadjusted) was a better
predictor of annual tree growth than non-integrated traits (WSR,
leaf size, LMA, canopy volume and tree height; i.e. question b), we
fit each trait separately for the above individual-level traits model
set (Table 2). In sum, we considered 135 possible model combina-
tions in total for this aspect of the study. We determined which
traits better described tree seasonal grow th based on the AICc
model selection. We also compared the variance explained by each
traits (fixed factors) using variance explained (R2
GLMM; Nakagawa &
Schielzeth, 2013). We fit the linear mixed-effects models use the
lmer function from the Lme4 package (Bates, Maechler, Bolker, &
Walker, 2015) in R 3.0.3 (R Development Core Team, 2014).
(6)
Z
i,t=𝛽0+𝛽1
VPD
t+𝛽2
trait
i+𝛽3
NCI
i+𝛽4
VPD
t×
trait
i+𝛽5
VPDt
×NCI
i
+
𝛽6
NCIH
i
+
𝛽7
NCIS
i
+
𝛽8
Season +
𝜏t
+
𝜑s
,
TABLE 2 Comparison of goodness-of-fit and model predictive
accuracy of non-integrative traits and integrative traits in growth–
traits models, growth–neighbourhood models and growth–climate–
neighbourhood models
Explanatory
variables R2
GLMM(m) R2
GLMM(c) AIC BIC
(a) Growth–traits models
LMAadjusted 2.42% 23% 4 3,645 43,700
LMA 0.05% 24% 43,809 43,864
Canopy volume 0.32% 24% 43,782 43,863
Height 2.13% 23% 43,6 49 43,70 8
WSR 0.77% 25% 4 3,769 43,823
Leaf area 0.40% 24% 43,815 43,870
(b) Growth–neighbourhood models
LMAadjusted 5.10% 23% 43,286 43 ,413
LMA 3.60% 23% 43,370 43,497
Canopy volume 4.90% 24% 43,334 43,4 61
Height 4.90% 24% 43,325 43,452
WSR 3.30% 24% 43,367 43,494
Leaf area 3.0 0% 24% 43,412 43,539
(c) Growth-climate-neighborhood models
LMAadjusted 7. 2 0 % 26% 43,059 43,167
LMA 4.40% 25% 43, 205 43,314
Canopy volume 5.00% 25% 4 3,175 43,283
Height 6.14% 26% 43,125 43, 234
WSR 4.28% 26% 43,183 43,292
Leaf area 4.45% 26% 43,221 43,330
Notes: (a) Growth–traits models:
Y
i
,
t=
𝛽0
+
𝛽1traits
i+
𝜄
t+
𝜑
s+
𝛾i
;
(b) Growth–neighbourhood models:
Y
i
,
t=𝛽
0
+𝛽
1traits
i+𝛽
2traits
i×
NCI
i
+
𝛽3traitsi
×
NCIHi
+
𝛽4traitsi
×
NCISi
+
𝜄t
+
𝜑s
+
𝛾i
; (c) Growth–climate
models:
Y
i
,
t=𝛽
0
+𝛽
1VPD
+𝛽
2VPD
×
traits
i+𝛽
3VPD
×
NCI
i+𝛽
4NCIH
i
+
𝛽5NCISi
+
𝜏t
+
𝜑s
+
𝛾i
.
The LMAadjusted represents LMA × crown volume, the WSR represents
wood-specific resistance. The R2
GLMM(m) is the marginal R2 which
variance explained by the fixed factors; the R2
GLMM(c) is conditional
R2 which variance explained by the entire model. AIC is the Akaike
information criterion value; BIC is the Bayesian information criterion.
The AIC and BIC values were calculated using ML and REML
estimations. The bold values represent statistically significant result s.

32
|
Journal of Ecology
YANG et Al.
3 | RESULTS
3.1 | Individual-level and species-level average trait
data and annual tree growth
This first set of analyses was designed to test whether mod-
els using individual-level trait data outperform the models using
species-level trait data. Overall, models that included intraspecific
trait variation had better support than models with interspecific
variation (Table 1). Both marginal and conditional R2
GLMM values
from species-level mean trait model are relatively low (3.8% and
5.9%) compared with R2
GLMM values from individual-level trait
model (10.3% and 12%; Table 1). Thus, the individual-level data
provided better predictions of annual growth. The R2
GLMM(m) val-
ues of the two models also indicated larger effects from the fixed
factors in the individual-level trait model than species-level mean
trait model. We also found the magnitude of trait effects on tree
growth varied depending on where individual- or species-level
trait was included (Figure 2). Specifically, WSR was not linked to
growth when using individual-level data, but was linked to growth
when aggregated as a species-level mean.
3.2 | Crown volume adjusted leaf traits and
tree growth
The best individual-level trait models for annual and seasonal
growth included the LMAadjusted values where larger values had
faster growth (Figures 2 and 4). The use of this integrative trait (i.e.
LMAadjusted) consistently generated superior models than those that
used non-integrative traits (e.g. leaf area, WSR). Specifically, the
best fit growth–traits models, growth–neighbourhood model and
growth–climate model all included LMAadjusted and no other traits
(Figure 2). The marginal and conditional R2
GLMM values from the in-
tegrative trait models were higher than other models ( Table 2) and
both the AIC and BIC values of integrative trait models are lower
than other non-integrative trait parameterized models ( Table 2).
3.3 | Interactive effects of traits, neighbourhoods
and climate on tree growth
As noted above, LMAadjusted was the only individual trait consist-
ently related to annual and seasonal grow th in this study (Figures 2
and 4). Larger LMAadjusted tree experienced faster growth than
trees with lower values. Neighbourhood effects had a strong im-
pact on in di vi dual tre e grow th, with greater neighbourhood crowd-
ing (i.e. NCI) resulting in lower individual growth rates both in the
individual-level and species mean trait models (Figures 2 and 4).
Neighbourhood crowding weighted by trait similarity (NCIS) had a
strong positive impact on annual tree grow th when using species
me an tr ait data (Fig ure 2b) . The VPD ha d a sig nif ica nt ef fect on tr ee
seasonal growth, where within season tree growth rate increased
when the VPD decreased (Figure 4b). However, there was no im-
pact of VPD on an nu al tr ee growth (Figure 4a). In ot her words , VPD
was related to tree growth within, but not across, seasons.
There were several important interactions identified both in our
annual and seasonal growth models. We found that LMAadjusted me-
diated the effect of increased neighbourhood crowding (i.e. NCI).
Specifically, as tree LMAadjusted increased, the negative effect of
crowding on growth increased (Figure 2a). The impact of LMAadjusted
on both tree annual and seasonal growth was mediated by VPD
(Figure 4). Specifically, as VPD increased, the positive effect of
LMAadjusted on growth intensified. However, an impact of VPD on
neighbourhood competition was not detected (Figure 4).
4 | DISCUSSION
We estimated how the LMA, tree height, WSR and leaf area of spe-
cies and local neighbourhood effects shape the observed response
in tree growth through a consideration of individual- and species-
level trait variation and seasonal drought effec ts on the growth of
36 tropical tree species over 8 years. The results of the study indi-
cate that individual-level trait information is a superior predictor of
growth as compared to species-level trait information. Furthermore,
FIGURE 2 Standardized regression
coefficients of the best-fitted model for
modelling the effects of the focal trait s,
neighbourhood crowding index (NCI),
neighbourhood hierarchical trait distance
(NCIH) and neighbourhood absolute trait
distance (NCIS) on tree annual growth.
Focal traits were either individual-level
(a) or average values at species-level (b).
Each point is a standardized regression
coefficient, and each line segments is
a 95% percentile interval respectively.
Filled circles represent significant effects.
LMAadjusted represents the LMA × crown
volume, WSR represents wood-specific
resistance
Individual-level trait values Species mean trait values
(a) (b)

|
33
Journal of Ecolog
y
YANG et Al.
growth is best predicted when LMA is placed in a whole plant alloca-
tion context and that growth is related to interactions between the
local neighbourhood biotic context and the dynamic climatic context
across seasons. In the following, we discuss the results in more detail.
4.1 | Intraspecific versus interspecific models
The treatment of individual heterogeneity in demography from func-
tional traits still remains largely unexplored (Yang et al., 2018). Here
we showed that individual-level trait data provide stronger models of
tree growth than species average trait values (Table 1; Figure 2). The
use of species-level average trait data in tree community ecology is
widespread (e.g. Kraft, Valencia, & Ackerly, 2008; Swenson, 2013).
While it is more practical to estimate a mean than measure traits on
every individual, a great deal of valuable information is likely lost in
suc h st ud ie s (A lb er t et al., 2010; Bol ni ck et al., 2011; Liu et al. , 2016).
Furthermore, the best fit individual-level models deviated in multi-
ple important ways from those models generated using species-level
mean trait data. For example, species-level models identified WSR
as being negatively associated with growth (Figure 2b). Not only was
this relationship not found in the individual-level models, it is the
opposite of the expected relationship between WSR, a correlate
of wood density and growth (Chave et al., 2009). In other words,
a model that aggregated trait data provided results that were in-
consistent with refined models using individual-level data and were
inconsistent with the expected relationships between a trait and
growth. Thus, the question arises—how reliable are the results from
previous studies that used species-level average traits for modelling
growth or other demographic rates? We cannot definitively answer
this question upon the basis of our single study, but it does suggest
that additional investigations similar to the present study are needed
to determine how large or small of an issue it is.
4.2 | Integrating leaf mass per area into a crown
volume context
In addition to the importance of individual-level trait data, we in-
tegrated a commonly measured trait, (i.e. LMA), into a whole tree
phenotypic context (i.e. LMAadjusted) in our trait-based models of
tree growth (e.g. Poorter, 1989). Our result s have shown that inte-
grating LMA with crown volume results in stronger models of tree
growth (Figure 2; Table 2). A considerable amount of research has
demonstrated important trade-offs described by commonly meas-
ured leaf, stem and propagule traits. However, these trait s are often
weakly correlated with tree demographic performance (e.g. Hérault
et al., 2011; Poorter et al., 20 08). The ability of traits to predict plant
performance remains largely unexplored (Shipley et al., 2016). Most
prior work has relied on using single trait, such as LMA, that lack
important contextual information about whole plant allocation and
other trait axes. In the case of trees, variation in resource allocation
is expected to be particularly important as species var y widely in
their allocation strategies and even within species as relative allo-
cation is not constant through ontogeny (e.g. Enquist et al., 2007).
The LMAadjusted variable in our study is developed from the exist-
ing, but less well-appreciated, functional ecology literature and is
based on the on the premise that total allocation to resource capture
should scale with individual growth (e.g. Enquist et al., 2007; Niklas
& Enquist, 2001; Poor ter, 1989). Our results showed that, indeed,
LMAadjusted was a better predictor of growth than other individual
functional traits (Figure 3). The better performance of integra-
tive traits in the predic tive models may help to explain why recent
FIGURE 3 Comparison of the standardized regression coefficients
of non-integrative trait s and integrative traits in three models fit to
the tree annual growth. The growth–traits model (a), the growth–
neighbourhood model (b) and the grow th–climate–neighbourhood
model (c). Each point is a standardized regression coefficient,
and each line segment is a 95% percentile interval respectively.
Filled circles represent significant effects. LMA represent the leaf
mass per area, LMAadjusted represents the LMA × crown volume,
WSR represents wood-specific resistance. Grow th–traits models:
Y
i
,
t=𝛽
0
+𝛽
1traits
i+𝜄t+𝜑s+𝛾
i
; growth–neighbourhood models:
Y
i
,
t=𝛽
0
+𝛽
1traits
i+𝛽
2traits
i×
NCI
i+𝛽
3traits
i×
NCIH
i+𝛽
4traits
i
×
NCISi
+𝜄
t
+𝜑
s
+𝛾
i
; growth–climate models:
Y
i
,
t=𝛽
0
+𝛽
1VPD +
𝛽2VPD
×
traitsi
+𝛽
3VPD
×
NCIi
+𝛽
4NCIHi
+𝛽
5NCISi
+𝜏
t
+𝜑
s
+𝛾
i
Leaf area
WSR
Height
LMA
Canopy volume
LMA
adjusted
–4 –2 024
Leaf area
WSR
Height
LMA
LMA
adjusted
Leaf area
WSR
Height
LMA
LMA
adjusted
(a)
(b)
(c)
Canopy volume
Canopy volume

34
|
Journal of Ecology
YANG et Al.
global analyses linking functional traits and demographical rates at
different sites have lacked high predictive ability (Adler, Fajardo,
Kleinhesselink, & Kraft, 2013; Poorter et al., 2008). Thus, future re-
search should continue to push towards the modelling of individual
demographic rates based upon individual traits and their phenotypic
context ( Yang et al., 2018).
4.3 | Traits mediate tree response to neighbourhood
competition
The importance of biotic interactions on an individual's per-
formance is well-appreciated in the forest ecology literature
(e.g. Canham et al., 2004, 2006). When considering sessile or-
ganisms, such as trees, the most evident biotic interactions
are expected to be restricted to proximate neighbours (Stoll &
Newbery, 2005). By combining spatially explicit demographic
censuses with individual-level functional traits, our results show
that neighbourhood crowding has consistently negative effects
on tree growth in this hyper-diverse tropical forest (Figure 2).
Specifically, our results show that simple neighbourhood crowd-
ing influences tree growth only when individual-level data were
utilized (Figures 2a and 4). This implies the effects of local-scale
processes such as competition may be harder to detect using
species mean data (Yang et al., 2018). It also means interspe-
cific trait variation may blur causal trait and environment inter-
action relationships (Yang et al., 2018). Furthermore, we show
that crowding effects are increased as crown allocation (i.e.
LMAadjusted) decreased in the focal individual and individuals
with higher canopy allocation had weaker negative effects due
to crowdi ng (F igu re 2a). Th us, th e s tro ng ne gat ive impa cts of le ss
crown allocation are noticeable at the individual-level as well as
in the context of the allocation of neighbouring individuals and
their density, which could not be detected with species-level
trait data (Figure 2b).
4.4 | Climate mediated growth–traits relationships
Previous studies have highlighted linkages between demography
and functional traits (e.g. Fortunel, Valencia, Wright, Garwood, &
Kraft, 2016; Liu et al., 2016; Poorter et al., 2008) or climatic factors
(e.g. Chmura et al., 2011; Clark et al., 2014; Johnson et al., 2018).
The relationship between traits and demographic performance are
typically modulated by the environment (Ellswor th & Reich, 1992).
However, most of this work has not directly aimed to uncover the
interacting effects of these two factors on tree demography. In the
final part of this study, we sought to quantify how functional traits
interact with climate to determine individual tree growth responses.
Specifically, the negative impact of VPD on growth was only evi-
dent within a season. A relationship between VPD and tree annual
growth was, however, not found. We expect this is due to total an-
nual tree growth being more strongly related to precipitation levels
totalled over the wet and dry seasons, whereas growth within dry
seasons should be related to variation in VPD. This may be expected
as strongly seasonal VPD should cause stomatal closure in response
to water deficit in the dry season, which limits to photosynthesis,
and constitutes a key cost in terms of plant growth (Mar tin-StPaul,
Delzon, & Cochard, 2017).
Beyond the VPD-growth relationship itself, we uncovered an im-
portant interaction between VPD and traits. The positive impact of
LMAadjusted on growth was mediated by VPD for both annual growth
and seasonal growth. High canopy allocation (i.e. high LMAadjusted)
was positively associated with both annual and seasonal growth.
Howeve r, the adva nta ge of hig h LMA adjusted was on ly re alize d as VPD
increased. In other words, trees with large total allocation to crown
leaf mass had better growth. This result may reflect the importance
of an unmeasured covariate of overall crown size. Specifically, larger
trees have larger crowns, but they also have larger root systems.
Thus, trees with larger crowns may be expected to perform better in
periodic dry conditions due to their larger root systems that will be
able to reach soil water in much lower strata.
FIGURE 4 Standardized regression
coefficients of the best-fitted model for
modelling individual trait effects, VPD,
neighbourhood effects (NCI), the effect of
hierarchical trait difference (NCIH) and the
effect of absolute trait difference (NCIS)
on tree grow th. VPD and tree growth
were either annual level (a) or seasonal
level (b). Each point is a standardized
regression coefficient, and each line
segment is a 95% percentile interval
respectively. Filled circles represent
significant effect. VPD represents the
vapour pressure deficit, LMAadjusted
represents the LMA × crown volume,
WSR represents wood-specific resistance
Individual annual growth Individual seasonal growth
(a) (b)

|
35
Journal of Ecolog
y
YANG et Al.
5 | CONCLUSIONS
Identif ying the drivers of tree growth is essential for our under-
standing of forest dynamics. Trait data are frequently used to iden-
tify these drivers, but trait–growth correlations are often weak
potentially due to a lack of contextual information regarding the
environment and whole plant allocation and an over-reliance on
species-level average trait values. Here we have shown that both
the climatic and local biotic contexts are important drivers of tree
demographic performance and have interactive effects such that
our understanding of the importance of one driver is faulty without
a consideration of the other. Furthermore, we show that individual-
level trait data generally provide stronger and alternative models of
tree growth as compared to models parameterized using species-
level average data. Combined, our results show that the relationships
between traits and tree growth are the outcomes of several contexts
and best studied at the individual level, which is generally not the
approach taken in trait-based tree community ecology. Thus, while
there is great promise in trait-based approaches to tree ecolog y, our
abilit y to un der sta nd th e dr ivers of popul at ion and commun it y struc-
ture and dynamics in the current and in future climates will be limited
if contextual and individual-level data remain understudied.
ACKNOWLEDGEMENTS
This research was supported by the Strategic Priorit y Research
Program of the Chinese Academy of Sciences (XDB31000000),
National Natural Science Foundation of China (31800353 and
31670442), the CAS 135 program (No. 2017X TBG-T01), the
Chinese Academy of Sciences Youth Innovation Promotion
Association (2016352), the Southeast Asia Biodiversity Research
Institute, Chinese Academy of Sciences (Y4ZK111B01), the West
Light Foundation of the Chinese Academy of Sciences and the ‘Ten
Thousand Talents Program of Yunnan’ (YNWR-QNBJ-2018-309).
N.G.S. was funded by NSF US-China Dimensions of Biodiversity
grants (DEB-1241136, DEB-1046113). We are grateful for support
from Xishuangbanna Station for Tropical Rain Forest Ecosystem
Studies (XSTRES). We thank Dr. Shirley Xiaobi Dong for assistance
with band installation and training. We thank Mr. Yongzheng Shen
an d Ms. Ye Liu fo r their ha rd wo r k in the fu n c t ion a l sampl e coll ect ion .
AUTHORS' CONTRIBUTIONS
J.Y., J. Z. and N.G.S. designed the study; J.Y., J.Z., Y.C. and X.S. per-
formed the analysis; J.Y., X.S., M.C., X.D., W.Z., X.Y. and G.Z. col-
lected the field data; J.Y., J.Z. and N.G.S. wrote the manuscript, and
all authors provided comments.
PEER REVIEW
The peer review history for this article is available at https://publo
ns. com/publo n /10.1111/1365-2745.13439
DATA AVAIL AB I LI T Y STATE MEN T
All data are available from the Dryad Digital Repository: https://doi.
org/10.5061/dryad.w9ghx 3fm5 (Yang et al., 2020).
ORCID
Ji e Ya ng https://orcid.org/0000-0002-4444-8240
Xiaoyang Song https://orcid.org/0000-0001-9529-1418
Min Cao https://orcid.org/0000-0002-4497-5841
Nathan G. Swenson https://orcid.org/0000-0003-3819-9767
REFERENCES
Adler, P. B., Fajardo, A., Kleinhesselink, A. R., & Kraft, N. J. B. (2013).
Trait-based test s of coexistence mechanisms. Ecology Letters, 16,
1294–1306. https://doi.org/10.1111/ele.12157
Alber t, C. H., Thuiller, W., Yoccoz, N. G., Douzet, R., Aubert, S., & Lavorel,
S. (2010). A multi-trait approach reveals the structure and the relative
importance of intra-vs. interspecific variability. Functional Ecology, 24(6),
1192–1201. ht tps://doi.org /10.1111 /j.1365-2435.2010 .01727.x
Arnold, S. J. (1983). Morphology per formance and fitness. American
Zoologist, 23, 347–361. https://doi.org/10.1093/icb/23.2.347
Bates, D., Maechler, M., Bolker, B., & Walker, S. (2015). lme4: Linear
mixed-effects mo dels using Eigen and S4. R package version 1.1-8.
Retrieved from http://CRAN.R-proje ct .org/packa ge=lme4
Bolnick , D. I., Amarasekare, P., Araújo, M. S., Bürger, R., Levine, J. M.,
Novak, M., … Vasseur, D. A. (2011). Why intraspecific trait variation
matters in community ecology. Trends in Ecolog y & Evolution, 26, 183–
192. https://doi.org/10.1016/j.tree.2011.01.009
Canham, C. D., Lepage, P. T., & Coates, K. D. (2004). A neighborhood
analysis of canopy tree competition: Effec ts of shading versus
crowding. Canadian Journal of Forest Research, 34, 778–787. https://
doi.org/10.1139/x03-232
Canham, C. D., Papaik, M., Uriarte, M., & McWilliams, W. (2006).
Neighborhood analyses of canopy tree competition along environ-
mental gradients in New England forests. Ecological Applications, 16,
540–554. https://doi.org/10.1890/1051-0761(2006)016[054 0:NA
OCT C]2.0.CO;2
Cao, M., Zhu, H., Wang, H., L an, G. Y., Hu, Y. H., Zhou, S. S., … Cui, J. Y.
(2008). Xishuangbanna tropical seasonal rainforest dynamics plot: Tree
distribution maps diameter tables and species documentation. Kunming,
China: Yunnan Science and Technology Press.
Cao, M., Zou, X., Warren, M., & Zhu, H. (20 06). Tropical forests of
Xishuangbanna China. Biotropica, 38, 306–309. https://doi.org/
10.2307/30043247
Chave, J., Coomes, D., Jansen, S., Lewis, S. L., Swenson, N. G., & Zanne, A. E.
(2009). Towards a worldwide wood economics spectrum. Ecology Letters,
12, 351–366. https://doi.org/10.1111/j.1461-0248.2009.01285.x
Chmura, D. J., Anderson, P. D., Howe, G. T., Harring ton, C. A., Halofsk y, J.
E., Peterson, D. L., … St.Clair, J. B. (2011). Forest responses to climate
change in the nor thwestern United States: Ecophysiological founda-
tions for adaptive management. Forest Ecology and Management, 261,
1121–1142. https://doi.org/10.1016/j.foreco.2010.12.040
Clark, J. S., Bell, D. M., Kwit, M. C., & Zhu, K. (2014). Competition-
interac tion landscapes for the joint response of forest s to climate
change. Global Change Biology, 20, 1979–1991 . ht tp s://doi.org /
10.1111/gcb.12425
Cornelissen, J. H. C., Lavorel, S., Garnier, E., Díaz, S., Buchmann, N.,
Gurvich, D. E., … Poorter, H. (2003). A handbook of protocols for
standardised and easy measurement of plant functional traits
worldwide. Australian Journal of Bota ny, 51, 335–380. https://doi.
org /10.1071/ bt02124
Eamus, D., Boulain, N., Cleverly, J., & Breshears, D. D. (2013). Global
change-type drought-induced tree mort ality: Vapor pressure deficit
is more important than temperature per se in causing decline in tree
health. Ecology and Evolutio n, 3, 2711–2729. https://doi.org/10.1002/
ece3.664
Ellswor th, D. S., & Reich, P. B. (1992). Leaf mass per area, nitrogen con-
tent and photosynthetic carbon gain in acer saccharum seedlings in

36
|
Journal of Ecology
YANG et Al.
contrasting forest light environments. Functional Ecology, 6, 423–
435. https://doi.org/10.2307/2389280
Enquist , B. J., Kerkhoff, A. J., Stark, S. C., Swenson, N. G., McCarthy, M.
C., & Price, C. A. (2007). A general integrative model for scaling plant
growth and functional trait spectra. Nature, 449, 218–222. https://
doi.org/10.1038/na tur e0 60 61
Enquist , B. J., & Lef fler, A. J. (2001). Long-term tree ring chronologies
from sympatric tropical dry forest trees: Individualistic responses to
climatic variation. Journal of Tropical Ecology, 17, 41–60. https://doi.
org /10.1017/S0266 46740 1001031
Ford, K. R., Bre ckh ei me r, I. K., Fra nk lin, J. F., Fre un d, J. A., Krois s, S. J., La rs on,
A. J., … HilleRisLambers, J. (2017). Competition alters tree grow th re-
sponses to climate at individual and stand scales. Canadian Journal of
Forest Research, 47, 53–62. https://doi.org/10.1139/cjfr-2016-0188
Fortunel, C., Valencia, R., Wright, S. J., Garwood, N. C., & K raft , N. J.
B. (2016). Functional trait differences influence neighbourhood in-
teractions in a hyperdiverse Amazonian forest. Ecology Letters, 19,
1062–1070. https://doi.org/10.1111/ele.12642
Grogan, J., & Schulze, M. (2012). The impact of annual and seasonal rain-
fall patterns on growth and phenology of emergent tree species in
Southeastern Amazonia Brazil. Biotropica, 44, 331–340. https://doi.
org /10.1111/j .1744 -7429.2011.00 825.x
Hérault, B., Bachelot, B., Poorter, L ., Rossi, V., Bongers, F., Chave, J., …
Baraloto, C . (2011). Func tional traits shape ontogenetic growth
trajec tories of rain forest tree species. Journal of Ecology, 99, 14 31–
144 0. https ://doi.or g/10.1111/j.1365-2745. 2011 .0188 3. x
Iida, Y., Kohyama, T. S., Swenson, N. G., Su, S.-H., Chen, C.-T., Chiang,
J.-M., & Sun, I.-F. (2014). Linking functional traits and demo-
graphic rates in a subtropical tree communit y: The importance
of size dependency. Journal of Ecology, 102, 641–650. https://doi.
org /10.1111/1365-2745 .12221
Iida, Y., Poor ter, L ., Sterck, F. J., Kassim, A . R ., Kubo, T., Pott s, M. D., &
Kohyama, T. S. (2012). Wood density explains architectural differ-
entiation across 145 co-occurring tropical tree species. Functional
Ecology, 26 , 274–282. htt ps://doi.or g/10 .1111/ j.13 65-24 35.2011.
01921 .x
Isik, F., & Li, B. (2003). Rapid assessment of wood density of live trees
using the Resistograph for selection in tree improvement programs.
Canadian Journal of Forest Research, 33, 2426–2435. https://doi.
org /10.1139/x0 3-176
Johnson, D. J., Needham, J., Xu, C., Massoud, E. C., Davies, S. J.,
Anderson-Teixeira, K . J., … McMahon, S. M. (2018). Climate sensitive
size-dependent survival in tropical trees. Nature Ecology & Evolution,
2, 1436. https://doi.org/10.1038/s4155 9-018-0626-z
J u n g , V. , V i o l l e, C . , M o n d y , C . , H o f f m an n , L . , & M u l l er , S. ( 2 01 0 ) . I n t r a sp e ci f ic
variability and trait-based community assembly. Journal of Ecology,
98, 113 4–1140. ht tp s://doi.o rg /10.1111/j.1365-2745. 2010.01687.x
Katabuchi, M. (2015). LeafArea: An R package for rapid digital image
analysis of leaf area. Ecological Research, 30, 1073–1077. htt ps://doi.
org /10.1007/s1128 4-015-1307-x
Kelly, A. E., & Goulden, M. L. (2008). Rapid shifts in plant distribution
with recent climate change. Proceedings of the National Academy of
Sciences of the United States of America, 105, 11823–11826. h tt ps://
doi.org/10.1073/pnas.0 8028 91105
Kraft, N. J. B., Valencia, R., & Ackerly, D. D. (2008). Functional traits
and niche-based tree community assembly in an Amazonian forest.
Science, 322, 580–582. https://doi.org/10.1126/scien ce.1160662
Kunstler, G., Lavergne, S., Courbaud, B., Thuiller, W., Vieilledent, G.,
Zimmermann, N. E., … Coomes, D. A. (2012). Competitive interac-
tions between forest trees are driven by species’ trait hierarchy,
not phylogenetic or functional similarity: Implications for forest
community assembly. Ecology Letters, 15, 831–840. https://doi.
org /10.1111/j .1461- 024 8. 2012.018 03.x
Lasky, J. R., Uriarte, M., Boukili, V. K., & Chazdon, R. L. (2014). Trait-
mediated assembly processes predict successional changes in
community diversity of tropical forests. Proceedings of the National
Academy of Sciences of the United States of America, 111, 5616–5621.
https ://doi.or g/10.1073/pnas.13193 42111
Liu, X., Swenson, N. G., Lin, D., Mi, X., Umaña, M. N., Schmid, B., & Ma, K.
P. (2016). Linking individual-level functional tr aits to tree growth in a
subtropical forest. Ecology, 97, 2396–2405. https://doi.org/10.1002/
ec y.14 45
Martin-StPaul, N., Delzon, S., & Cochard, H. (2017). Plant resistance to
drought depends on timely stomatal closure. Ecology Letters, 20,
1437–1447. ht tps://doi .org /10.1111 /ele .12851
McDowell, N., Pockman, W. T., Allen, C. D., Breshears, D. D., Cobb, N.,
Kolb, T., … Yepez, E. A. (2008). Mechanisms of plant survival and
mortality during drought: Why do some plants survive while oth-
ers succumb to drought? New Phytologist, 178, 719–739. https://doi.
org /10.1111/j .1469-8137.20 08.02436.x
McGill, B. J., Enquist, B. J., Weiher, E., & Westoby, M. (2006).
Rebuilding community ecology from functional trait s. Tren ds in
Ecology & Evolution, 21, 178–185. https://doi.org/10.1016/j.tree.
2006.02.002
McKenney, D. W., Pedlar, J. H., Lawrence, K., Campbell, K., & Hutchinson,
M. F. (2007). Potential impacts of climate change on the distribu-
tion of North American trees. BioScience, 57, 939–948. https://doi.
org /10.1641/B571106
Miles, L., Newton, A. C., DeFries, R . S., Ravilious, C., May, I., Blyth, S., …
Gordon, J. E. (20 06). A global overview of the conservation status of
tropical dry forest s. Journa l of Biogeography, 33, 491–505. ht tps: //do i.
org /10.1111/j .136 5-2699.20 05.01424. x
Moles, A . T. (2018). Being John Harper: Using evolutionary ideas to
improve understanding of global patterns in plant trait s. Journal of
Ecology, 106 , 1–18. https://doi.org /10.1111/1365-2745.12887
Muller-Landau, H. C., & Dong, S. X. (2008). Metal band d endrom eter
protoco l. Version: November 2008. Retrieved from http://www.
ctfs.si.edu/data///docum ents/Metal_Band_Dendr ometer_Proto
col_20081 119_archi ve.pdf
Myneni, R. B., Yang, W., Nemani, R. R., Huete, A. R., Dickinson, R. E.,
Knyazikhin, Y., … Salomonson, V. V. (2007). Large seasonal swings in
leaf area of Amazon rainforests. Proceedings of the National Academy
of Science s of the United States of America, 10 4, 4820–4823. https://
doi.org/10.1073/pnas.0 6113 38104
Nakagawa, S., & Schielzeth, H. (2013). A general and simple method
for obtaining R2 from generalized linear mixed-ef fect s mod-
els. Methods of Ecology and Evolution, 4, 133–142. ht tps://doi.
org /10.1111/J.2041-210x.2012 .0 0261.X
Newber y, D. M., & Stoll, P. (2013). Relaxation of species-specific neigh-
borhood ef fect s in Bornean rain fores t under climatic perturbation.
Ecology, 94, 2838–2851. https://doi.org/10.1890/13-0366.1
Niklas, K. J., & Enquist, B. J. (20 01). Invariant scaling relationships for in-
terspecific plant biomass production rates and body size. Proceedings
of the Nation al Acade my of Sciences of the United Sates of America, 98,
2922–2927. https://doi.org/10.1073/pnas.04159 0298
Nowacki, G. J., & Abrams, M. D. (2015). Is climate an important driver
of post-European vegetation change in the Eastern United States?
Global Change Biology, 21, 314–33 4. htt ps://doi. org/10 .1111/
gcb.12663
O'Brien, M. J., Reynolds, G., Ong, R., & Hector, A. (2017). Resistance of
tropical seedlings to drought is mediated by neighbourhood diversity.
Nature Ecology & Evolutio n, 1, 1643–1648. https://doi.org/10.1038/
s4155 9-017-0326-0
Osnas, J. L. D., Lichstein, J. W., Reich, P. B., & Pacala, S. W. (2013). Global
leaf trait relationships: Mass area and the leaf economic s spectrum.
Science, 340, 741–744. https://doi.org/10.1126/scien ce.1231574
Paine, C. E. T., Amissah, L., Auge, H., Baraloto, C., Baruffol, M., Bourland,
N., … Hector, A. (2015). Globally functional trait s are weak predictors
of juvenile tree growth, and we do not know why. Journal of Ecology,
103, 978–989. https://doi.org/10.1111/1365-2745.12401

|
37
Journal of Ecolog
y
YANG et Al.
Pederson, N., D'Amato, A. W., Dyer, J. M., Foster, D. R., Goldblum, D.,
Hart, J. L., … Williams, J. W. (2015). Climate remains an import-
ant driver of post-European veget ation change in the eastern
United States. Global Change Biology, 21, 2105–2110. https://doi.
org /10.1111/gcb.12779
Poorter, H. (1989). Interspecific variation in relative growth rate: On
ecological causes and physiological consequences. In H. Lambers,
M. L. Cambridge, H. Konings, & T. L. Pons (Eds.), Causes and con-
sequences of variation in growth rate and productivity of higher
plants (pp. 45–68). The Haguehe, The Netherlands: SPB Ac ademic
Publishing.
Poorter, H ., Niinemets, U., Poorter, L ., Wright, I. J., & Villar, R . (2009).
Causes and consequences of variation in leaf mass per area (LMA):
A meta-analysis. New Phytologist, 182, 565–588. https://doi.
org /10.1111/j .1469-8137.20 09.028 30 .x
Poorter, L., Wright, S. J., Paz, H., Ackerly, D. D., Condit, R., Ibarra-
Manríquez, G., … Wright, I. J. (2008). Are functional traits good
predictors of demographic rates? Evidence from five neotropical
forest s. Ecology, 89, 1908–1920. https://doi.org/10.1890/07-0207.1
Pretzsch, H., Biber, P., Schütze, G., Uhl, E., & Rötzer, T. (2014). Forest stand
growth dynamics in Central Europe have accelerated since 1870.
Nature Communications, 5, 4967. https://doi.org/10.1038/ncomm s5967
R Development Core Team. (2014). R: A language and environment for
statistical computing. Vienna, Austria: R Foundation for Statistical
Computing. Retrieved from http://www.R-proje c t.org/
Reich, P. B., Ellsworth, D. S., Walters, M. B., Vose, J. M., Gresham, C.,
Volin, J. C ., & Bowman, W. D. (1999). Generality of leaf trait relation-
ships: A test across six biomes. Ecology, 80, 1955–1969. https://doi.
org /10. 23 07/176671
Reich, P. B., Walters, M. B., & Ellsworth, D. S. (1997). From tropics to
tundra: Global convergence in plant functioning. Proceedings of the
National Academy of Sciences of the United States of America, 94,
13730–13734. https://doi.org/10.1073/pnas.94.25.13730
Restrepo-Coupe, N., da Rocha, H. R., Hutyra, L. R., da Araujo, A. C.,
Borma, L. S., Christoffersen, B., … Saleska, S. R. (2013). What drives
the seasonalit y of photosynthesis across the Amazon basin? A cross-
site analysis of eddy flux tower measurements from the Brasil flux
network. Agricultural and Forest Meteorology, 182, 128–144. https://
doi.org/10.1016/j.agrfo rmet.2013.04.031
Shipley, B., De Bello, F., Cornelissen, J. H. C., Laliber té, E., Laughlin,
D. C., & Reich, P. B. (2016). Reinforcing loose foundation stones in
trait-based plant ecology. Oecologia, 180 , 923–931. https://doi.
org /10.1007/s00 44 2-016-35 49-x
Stoll, P., & Newbery, D. M. (2005). Evidence of species-specific neigh-
borhood effects in the Dipterocarpaceae of a Bornean rain forest.
Ecology, 86, 3048–3062. https://doi.org/10.1890/04-1540
Swenson, N. G. (2013). The assembly of tropical tree communities-the
advances and shortcomings of phylogenetic and functional trait anal-
yses. Ecography, 36, 264–276. https://doi.org/10.1111/j.1600-0587.
2012.00121.x
Swenson, N. G., Stegen, J. C., Davies, S. J., Erickson, D. L., Forero-
Montaña, J., Hurlbert, A. H., … Zimmerman, J. K. (2012). Temporal
turnover in the composition of tropical tree communities: Func tional
determinism and phylogenetic stochasticity. Ecolog y, 93, 490–499.
https://doi.org/10.1890/11-1180.1
Uri ar te , M. , Lask y, J. R., Bouk il i, V. K., & Chazdo n, R. L . (2016). A trait-me-
diated neighbourhood approach to quantify climate impact s on suc-
cessional dynamics of tropical rainforests. Functional Ecology, 30,
157–167. https://doi.org /10.1111/1365-24 35.12576
Uriarte, M., Muscarella, R., & Zimmerman, J. K. (2018). Environmental
heterogeneity and biotic interactions mediate climate impacts on
tropical forest regeneration. Global Change Biology, 24, e692–e704.
https ://doi.or g/10.1111/gcb.140 00
Uriarte, M., Swenson, N. G., Chazdon, R. L., Comita, L. S., John Kress, W.,
Erickson, D., … Thompson, J. (2010). Trait similarity shared ances-
try and the structure of neighbourhood interactions in a subtropical
wet forest: Implications for community assembly. Ecology Letters, 13,
1503–1514. https://doi.org/10.1111/j.1461-0248.2010.01541.x
Violle, C., Enquist , B. J., McGill, B. J., Jiang, L., Alber t, C. H., Hulshof, C., …
Messier, J. (2012). The return of the variance: Intraspecific variabil-
ity in communit y ecology. Trends in Ecology & Evolution, 27, 244–252.
https://doi.org/10.1016/j.tree.2011.11.014
von Oheimb, G., Lang, A. C., Bruelheide, H., Forrester, D. I., Wasche, I.,
Yu, M. J., & Härdtlea, W. (2011). Individual-tree radial growth in a
subtropical broad-leaved forest: The role of local neighbourhood
competition. Forest Ecology and Management, 261, 499–507. https://
doi.org/10.1016/j.foreco.2010.10.035
Westoby, M., & Wright, I. J. (2006). Land–plant ecology on the basis of
functional traits. Trends in Ecology & Evolution, 21, 261–268. https://
doi.org/10.1016/j.tree.2006.02.004
Will, R. E., Wilson, S. M., Zou, C. B., & Hennessey, T. C. (2013). Increased
vapor pressure deficit due to higher temperature leads to greater
transpiration and faster mortality during drought for tree seedlings
common to the forest-grassland ecotone. New Phytologist, 200, 36 6–
374. https: //doi.org/10 .1111/nph.12321
Wright, I. J., Reich, P. B., Westoby, M., Ackerly, D. D., Baruch, Z., Bongers,
F., … Villar, R. (2004). The worldwide leaf economics spectrum.
Nature, 428, 821–827. https://doi.org/10.1038/natur e02403
Wright, S. J., Kitajima, K ., Kraft, N. J. B., Reich, P. B., Wright, I. J., Bunker,
D. E., … Zanne, A . E. (2010). Functional traits and the growth-mor-
tality trade-off in tropical trees. Ecolog y, 91, 3664–3674. https://doi.
org /10.189 0/09-23 35.1
Yang, J., C ao, M., & Swenson, N. G. (2018). Why functional traits do not
predict tree demographic rates. Trends in Ecology & Evolution, 33,
326–336. https://doi.org/10.1016/j.tree.2018.03.003
Yang, J., Zhang, G., Ci, X., Swenson, N. G., Cao, M., Sha, L . Q., … Lin, L. X.
(2014). Functional and phylogenetic assembly in a Chinese tropical
tree community across size class spatial scales and habitats. Functional
Ecology, 28, 520–529. https://doi.org/10.1111/1365-2435.12176
Zambrano, J., Marchand, P., & Swenson, N. G. (2017). Local neighbour-
hood and regional climatic contexts interact to explain tree perfor-
mance. Proceedings of the Royal Societ y B: Biological Sciences, 284,
20170 523.
Ze ppel , M. J. B ., Wil ks, J. V., & Lew is, J. D. (20 14). Im pacts of ex tre me pre cip-
itation and seasonal changes in precipitation on plants. Biogeosciences,
11, 3083–3093. https://doi.org/10.5194/bg-11-3083-2014
Yang, J., Song, X ., Zambrano, J., Chen, Y., Cao, M., Deng, X., … Swenson,
N. G. (2020). Data from: Intraspecific variation in tree growth re-
sponses to neighbourhood composition and seasonal drought in a
tropical forest. Dryad Digital Repository, https://doi.org/10.5061/
dryad.w9ghx 3fm5
SUPPORTING INFORMATION
Additional supporting information may be found online in the
Supporting Information section.
How to cite this article: Yang J, Song X, Zambrano J, et al.
Intraspecific variation in tree growth responses to
neighbourhood composition and seasonal drought in a tropical
forest. J Ecol. 2021;109:26–37. https://doi.org/10 .1111/1365-
2745.13439