Biological Conservation 161 (2013) 118–127

Contents lists available at SciVerse ScienceDirect

Biological Conservation
journal homepage: www.elsevier.com/locate/biocon

Use of provisioning ecosystem services drives loss of functional traits

across land use intensification gradients in tropical forests
in Madagascar
Kerry A. Brown a,⇑, Steig E. Johnson b, Katherine E. Parks c, Sheila M. Holmes b, Tonisoa Ivoandry d,
Nicola K. Abram e, Kira E. Delmore f, Reza Ludovic g, Hubert E. Andriamaharoa g,
Tracy M. Wyman b, Patricia C. Wright h
a
School of Geography, Geology and the Environment, Centre for Earth and Environmental Science Research (CEESR), Kingston University, Surrey KT1 2EE, UK
b
Department of Anthropology, University of Calgary, Calgary, Alberta, Canada
c
Centre for Environmental Sciences, University of Southampton, UK
d
Department of Plant Biology and Ecology, University of Antananarivo, Faculty of Sciences, Antananarivo BP 906, Madagascar
e
Durrell Institute of Conservation and Ecology (DICE), University of Kent Canterbury, Kent, UK
f
Department of Zoology, University of British Columbia, British Columbia, Canada
g
Missouri Botanical Garden, Antananarivo BP 906, Madagascar
h
Institute for the Conservation of Tropical Environments, Stony Brook University, Stony Brook, NY, United States

a r t i c l e i n f o a b s t r a c t

Article history: Ecosystems services are threatened by the rapid degradation of tropical rainforests. In light of these
Received 29 October 2012 threats, questions remain about how societies on the forest fringe who depend on provisioning services
Received in revised form 1 March 2013 affect plant functional traits. This study assessed the relationship between plant functional traits, forest-
Accepted 11 March 2013
dependent societies and provisioning ecosystem services along a forest–agriculture matrix gradient in
Available online 21 April 2013
tropical humid forests of Madagascar. Data were collected for six functional traits and six provisioning
services. We evaluated functional diversity (FD) and utilitarian diversity (FD of provisioning services)
Keywords:
along disturbance gradients. We also determined the traits most susceptible to loss along the gradient
Biodiversity
Functional diversity
as well as the relationship between functional traits and provisioning services. The results showed that
Functional traits FD, utilitarian diversity and species richness decreased significantly across all modified habitats. There
Ecosystem services was a distinct suite of traits absent from intensely used habitats and those same traits were associated
Land use intensification with the two most important provisioning services in the region (firewood and construction). The activ-
Madagascar ities of people living on the forest fringe seem to be a strong selective force on trait loss in plant commu-
Subsistence agriculture nities, which in turn will influence future species assembly and trait diversity and distribution. Moreover,
this study suggests that it is possible to predict trait loss from plant communities in forests where deg-
radation is mediated primarily through subsistence agriculture and resource extraction. Conservation
efforts should recognize that forest degradation and deforestation are contextually specific, determined
mostly by people’s efforts to maintain their basic livelihood, and therefore necessitate local-scale inter-
ventions that feed into landscape-scale policy initiatives.
Ó 2013 Elsevier Ltd. Open access under CC BY-NC-ND license.

1. Introduction locally, regionally and globally (Cardinale et al., 2012; Foley

Tropical forests house two-thirds of the world’s terrestrial bio-
diversity (Whitmore, 1998; Wright, 2002), providing essential cul-
tural, provisioning, regulating and supporting ecosystem services
⇑ Corresponding author. Tel.: +44 0208 417 2009.
E-mail addresses: K.brown@kingston.ac.uk (K.A. Brown), steig.johnson@ucalgar-
y.ca (S.E. Johnson), k.e.parks@soton.ac.uk (K.E. Parks), holmess@ucalgary.ca (S.M.
Holmes), tonisoaivoandry@yahoo.fr (T. Ivoandry), nka4@kent.ac.uk (N.K. Abram),
kdelmore@zoology.ubc.ca (K.E. Delmore), Reza.ludovic@mobot-mg.org (R. Ludovic),
ahubertemilien@yahoo.fr (H.E. Andriamaharoa), twyman@ucalgary.ca (T.M.
Wyman), patchapplewright@gmail.com (P.C. Wright).
et al., 2007; Kremen, 2005). As a consequence of factors ranging
from large-scale forest conversion to over-harvesting of resources
for subsistence, tropical forests continue to be affected by practices
that promote deforestation and forest degradation (Achard et al.,
2004; Asner et al., 2009). This has led to large areas of the tropics
being transformed into simplified, low diversity ecosystems char-
acterised by a forest–agriculture mosaic that ranges from mature
and secondary forests to agro-forestry plantations, arable crops,
pasture and nutrient-depleted fallow grass and scrubland (Gardner
et al., 2009; Morris, 2010). This transformation of natural land-
scapes and intensification of land use is often accompanied by

0006-3207 Ó 2013 Elsevier Ltd. Open access under CC BY-NC-ND license.

http://dx.doi.org/10.1016/j.biocon.2013.03.014
 K.A. Brown et al. / Biological Conservation 161 (2013) 118–127
reduced plant diversity (Kessler et al., 2009; Lin et al., 2011; Pardi-
ni et al., 2009). In addition to changes to species assemblages, these
dynamics may affect associated functional traits and delivery of
ecosystem services (Norris et al., 2010; Quetier et al., 2007), partic-
ularly provisioning services vital for the livelihoods of human com-
munities on the forest fringe (Brown et al., 2011; Cardinale et al.,
2012).
Two important questions regarding the interaction of land use
and tropical forest biodiversity are: (i) how forest-dependent soci-
eties affect plant functional traits and (ii) whether functional trait
loss is related to use of particular provisioning services. Functional
traits have been shown to link organisms, ecosystem properties
and ecosystem services (Cornwell et al., 2008; de Bello et al.,
2010; Violle et al., 2007). Traits are the ecological attributes of
organisms, which affect ecosystem properties, in turn modulating
ecosystem services (Lavorel et al., 2011; Naeem et al., 2012). Bio-
logical communities with high functional trait diversity and redun-
dancy are usually more resilient, better buffered against the
vagaries of environmental change (Laliberté et al., 2010; McGill
et al., 2010), and are generally associated with increased ecosystem
function (Flynn et al., 2011) and the ability to deliver ecosystem
services (Mouillot et al., 2011). A commonly assumed effect of land
use change on biodiversity suggests that as land use gradients
intensify, there should be a concomitant loss of species and func-
tional trait diversity (Srivastava and Vellend, 2005). Although
experimental studies have supported this relationship, the evi-
dence in natural communities has been equivocal. For instance,
land use intensification may lead to decreased species richness
and functional diversity for mammal and bird communities, but
not always for plant communities (Flynn et al., 2009; Laliberté
et al., 2010; Mayfield et al., 2005). Furthermore, species’ response
trajectories to land use change does not necessarily decrease spe-
cies richness and functional diversity, and indeed has the potential
to increase both (Mayfield et al., 2010). These observations suggest
that the relationship between species richness and functional
diversity is complex.
In this paper we focus on human-mediated degradation in the
humid tropical forests of southeastern Madagascar. Madagascar
is considered an area of high priority for conservation efforts and
is one of 34 global biodiversity hotspots (Mittermeier et al.,
2005), indicating a high level of plant endemism, but also high
rates of deforestation and degradation (Myers et al., 2000). Approx-
imately 80% of Madagascar’s population live in rural areas and de-
pend on ecosystem services from forests (Ingram and Dawson,
2006), directly influencing biodiversity through conversion of for-
estlands for crop cultivation, selective logging and extraction of
natural resources. These activities also affect vital ecosystem ser-
vices on which local people depend. Rather than a rapid loss of for-
ests, this process converts intact forests to degraded or agricultural
lands and has been the reason behind Madagascar’s historically
low deforestation rates (Green and Sussman, 1990; Harper et al.,
2007). With increasing population density (both urban and rural)
and changes in socio-economic drivers, this past trend is changing
and forests in Madagascar are becoming increasingly vulnerable to
large-scale forest clearance (Agarwal et al., 2005). Given the links
between functional traits and ecosystem services (and their poten-
tial degradation), it is important to determine if anthropogenic
activities have modified trait structure in forests.
This study has three primary objectives. First, we evaluate how
functional trait diversity varies according to disturbance both
along a forest–agriculture matrix gradient and between continuous
and fragmented environments. We focus on utilitarian plants –
wild plants that are extracted for firewood, charcoal, timber and
other non-timber forest products (NTFPs). We hypothesize that de-
clines in biodiversity will equate to loss of functionally important
plant traits within landscapes dominated by resource extraction.
119
Second, we determine which traits are most susceptible to loss
along disturbance gradients. We expect that functional traits asso-
ciated with specific, common provisioning services (e.g., firewood)
will be disproportionately lost from the most intensively modified
habitats. Finally, we determine the relationship between functional
traits and provisioning services used by subsistence communities.
For example, plants commonly used for construction are expected
to have high wood density—associated with strong timber—and
species with high values for that trait should be preferentially lost
from intensively modified habitats. Additionally, bark thickness
and wood density may both be important traits on which local
people implicitly focus when choosing species for fuelwood. The
extent to which resource extraction and subsistence agriculture
influence plant traits have not been fully explored in a tropical
setting.
2. Methods
2.1. Classifying disturbance gradients: forest–agriculture matrix and
continuous vs. fragmented forests
The vegetation datasets were assembled from published and
unpublished studies in southeastern regions of Madagascar
(Fig. 1). Twenty-six sites, totalling 235 plots were sampled for this
study. These studies were undertaken in areas with varying levels
of land use intensity. Many of the forests were likely in a state of
active degradation and almost certainly exhibited fluctuating de-
grees of disturbance over time, a state common to many habitats
in Madagascar (Irwin et al., 2010).
We classified vegetation plots from each study according to
their position on a forest–agriculture matrix gradient, as well as
whether they were found in continuous or fragmented forests
(Fig. 1). The forest–agriculture matrix gradient was divided into
five categories: (i) interior – entirely within forest; (ii) interior–
edge – combination of forest and edge habitats; (iii) interior–
edge-matrix – combination of interior forests, edge and matrix
habitats; (iv) edge-matrix – combination of edge and matrix habi-
tats; and (v) matrix – entirely within non-forest. Forest and non-
forest classes were obtained from layers created by Conservation
International (Harper et al., 2007), with additional imagery from
GoogleEarth 2012 and GeoEye 2011 used to correct for cloud cov-
ered areas. Next, a buffer 100 m from the forest edge was created
within forested areas. This buffer distance has been shown to
delineate marked differences in important tree characteristics,
such as variability in mortality (Laurance et al., 2007). Each plot
was then coded for their location according to where they fell in
the landscape in the five categories outlined above. It is important
to note that land use in Madagascar is dynamic, which could result
in errors in the categories assigned to plots (e.g., through defores-
tation or forest restoration between when the Conservation Inter-
national layer was classified and when the plots were inventoried).
Accordingly, we verified that the categories attributed to each plot
from the Conservation International forest cover layer did not
change when visually inspected using more recent and higher-res-
olution imagery from GoogleEarth and GeoEye; no such changes
were observed. Additionally, many of the sites used in this study
consisted of multiple plots with several different intensification
categories.
Forests were classified as continuous or fragmented using lay-
ers and satellite imagery indicated above. Vegetation plots desig-
nated as forest fragments were visibly isolated from other tracts
of forests, with distance between fragments ranging from 61 m
to 636 m. The area of natural forest fragments sampled ranged
from 73 ha to 1063 ha. Fragments vary in their properties (e.g., size
and isolation) and the distinction between large fragments and
120 K.A. Brown et al. / Biological Conservation 161 (2013) 118–127

Fig. 1. Site map showing study sites in southeastern Madagascar. Indicated are the number and size of plots at each site. aBrown et al. (2011), 49 plots, 0.16 ha; bBrown et al.
(2009), 16 plots, 0.04 ha; cHolmes (2012), 40 plots, 0.04 ha; dDelmore (2009), 72 plots, 0.02 ha; eJohnson (2002), 10 plots, 0.04 ha; fIngraldi (2010), Ludovic and
Andriamaharoa (unpublished data), 48 plots, 0.04 ha.

continuous forest is somewhat arbitrary. However, eastern Mada- divided by the few rivers that traverse the entire corridor). This
gascar has a prominent forest corridor that extends largely unbro- contrasts sharply with the forest remnants in the coastal plain,
ken along the entire eastern escarpment. Forest plots sampled in which can be characterized as relatively small forest fragment net-
this corridor maintain connectivity across the region (except when works that are well separated from each other.
 K.A. Brown et al. / Biological Conservation 161 (2013) 118–127
2.2. Functional traits
Plant traits were collected from forests within the boundaries of
Ranomafana National Park (RNP) and habitats surrounding the
park (Fig. 1). Individuals were selected from different habitats in
RNP, covering a large range of species occurrences. Trait sampling
was conducted from June 01 to June 21, 2011 and collection times
were restricted to 0800 and 1600 h each day. Plant functional traits
were sampled for 375 individuals across 75 species (i.e., five indi-
viduals per species). Species selected for trait measurements were
the most abundant utilitarian plants within the region (see Sec-
tion 2.6). Data were collected for six traits: WSG, bark thickness
(mm), specific leaf area (SLA, mm2 g 1), leaf size (mm2), leaf water
content (%) and height (cm). We selected traits that were regarded
as essential for plant strategy, but also have important associations
with provisioning services.
2.3. Wood specific gravity (WSG), bark thickness and plant height
WSG was measured using methods described in Cornelissen
et al. (2003), Muller-Landau (2004) and Williamson and Wiemann
(2010). This trait—also referred to as wood density (but see Wil-
liamson and Wiemann, 2010 for a distinction)—is defined as the
density of wood relative to the density of water and is calculated
by dividing the oven dried weight of wood by its fresh volume.
For this study, wood samples were extracted at 1.5 m from tree
base, perpendicular to the bark, using a 5.15 mm diameter incre-
ment borer to obtain cores 10–50 cm long. Samples were pro-
cessed within 5 h of field collection and within 2 h after
returning to the laboratory. Cores were divided into two or more
segments (cutting parallel to the bark) and the length of each
was measured. The water displacement method was used to mea-
sure the fresh volume of each core segment (Cornelissen et al.,
2003). After which, the samples were oven dried at 100 °C for
48 h or until a constant weight was achieved and measured using
a balance scale (accurate to within ±0.001 g). We calculated the
area-weighted mean of each core segment to account for the rela-
tive over-representation of wood toward the pith, in pith to bark
cores (Williamson and Wiemann, 2010). WSG of each core segment
was calculated as the oven dried weight divided by the fresh vol-
ume and is a unit-less value (Williamson and Wiemann, 2010).
The bark was kept intact on each core and bark thickness was mea-
sured after measuring fresh volume and prior to oven drying. WSG
and bark thickness were calculated as the average of five
measurements.
Tree height was either measured with a laser hypsometer when
feasible (dense understory can compromise visibility) or estimated
with reference to an adjacent measured tree. Height was calculated
as the average of 20 measurements.
2.4. Leaf traits: specific leaf area (SLA), leaf size and leaf water content
SLA was calculated as the ratio of leaf area (mm2) to leaf dry
mass (g) following Cornelissen et al. (2003). We collected five
leaves from each individual, limiting collection to leaves exposed
to full sunlight when possible and from the top of canopy for
understory trees. Immediately after collection, leaves were sealed
in plastic bags and stored at a relatively constant temperature until
they could be returned to the laboratory. All samples were pro-
cessed within 24 h of collection and those left overnight were
placed in moistened paper towels and stored in a refrigerator at
5 °C until morning when they could be processed. Fresh leaf mass
was measured using a balance (accurate to within ±0.001 g) and
leaf area was measured in the laboratory using a LI-3000C portable
area meter (LI-COR, Lincoln, Nebraska, USA). Leaf size (mm2) in-
cluded only the lamina, while the leaf area used to calculate SLA in-
121
cluded lamina, petiole and petiolules. The constituents from each
individual leaf (i.e., lamina, petiole and petiolules) were subse-
quently oven dried at 70 °C for 48–72 h. Leaf water content (%)
was calculated using all leaf constituents as the difference between
fresh leaf mass and oven dried leaf mass. SLA, leaf water content
and leaf size were calculated as the average of five measurements.
2.5. Functional diversity: Petchey and Gaston’s FD
We assess trait diversity using Petchey and Gaston’s FD (2002),
a multi-taxa index that measures the variation of traits for species
present in an ecological community and provides a way to com-
pare the diversity of species’ roles between plant communities.
Petchey and Gaston’s FD is a dendrogram-based index, originally
based on the phylogenetic diversity metrics, PD (Faith, 1992). We
used the Gower method to calculate multivariate distances be-
tween species and unweighted pair-group method with arithmetic
means (UPGMA) to summarize these pair-wise distance measures
into functional dendrograms. Differences between forest–agricul-
ture matrix gradient and continuous vs. fragmented forests in
terms of FD were evaluated using nonparametric tests, Kruskal–
Wallis and Mann–Whitney, respectively. The same analyses were
performed to assess differences in species richness. Post hoc
pairwise comparisons between land use categories using Mann–
Whitney tests with Bonferroni corrections were used to evaluate
significant Kruskal–Wallis results.
2.6. Utilitarian diversity (UD) and provisioning services
The utilitarian diversity index—similar to FD index except the
former determines the diversity of provisioning services (see
Brown et al., 2011)—was based on a species’ utility to residents
surrounding the study areas. Species’ uses of interest were as fol-
lows: construction (ordinal), firewood (ordinal), medicinal (bin-
ary), food (binary), wood for tools (binary), and wood for
furniture (binary). The utility for 205 species was established using
two approaches: (i) questionnaires and (ii) interviews with local
experts. The traditional ecological knowledge (TEK) of Malagasy re-
search technicians was employed to identify ecosystem service
providers (ESP, Luck et al., 2009), plants essential for maintaining
people’s livelihood. Secondly, we used data from structured house-
hold questionnaires that determined utility (for firewood and con-
struction timber) and preference values (for construction timber)
for plant species used by villages associated with Manombo Forest
(Abram, 2008; Brown et al., 2011). Construction materials, fuel-
wood, charcoal and non-timber forest products (NTFPs) are consid-
ered the most prominent selective uses of forests in Madagascar
(Consiglio et al., 2006) and there is intensive extraction of those re-
sources (Kremen et al., 1998). Of the 205 species for which we have
utilitarian information, we included only 75 in the utilitarian
diversity analyses, because we were constrained by the number
species for which functional traits were collected. Differences be-
tween forest–agriculture matrix gradient and continuous vs. frag-
mented forests in terms of utilitarian diversity were evaluated
using the same nonparametric approach as FD and species
richness.
2.7. Classification and regression tree
A classification and regression tree (CART) approach was used
to determine which traits were most likely to be lost from the ma-
trix category of the forest–agriculture gradient as well as frag-
mented forests. For the forest–agriculture gradient, we identified
species present in interior forests and absent from all other forest
classifications. For fragmentation, we identified all species present
in the continuous forests but absent from the forest fragments. The
122 K.A. Brown et al. / Biological Conservation 161 (2013) 118–127
CART analysis recursively partitioned the predictor variables
(traits) to explain the variability in the response variable, which
was whether each species was present in at least one of the
unmodified habitats, but none of the modified habitats. This ap-
plied to both types of disturbance regimes. We used the sum of
squares around group means to establish splitting criteria, selected
optimal tree size by K-fold cross-validation, and evaluated model
fit with Pearson correlation of predicted to observed species loss
(De’ath and Fabricius, 2000).
2.8. Multiple logistic regression
We performed a multiple logistic regression to determine the
relationship between provisioning services and functional traits.
We focused on two provisioning services, construction (timber)
and firewood, both of which are known to be important for local
residents in this region (Brown et al., 2011, 2009). The logistic
regression model estimated the probability that a species was used
for construction, firewood or both. We assigned ‘‘1’’ to every spe-
cies encountered in the vegetation surveys known to be used either
for construction (timber) or firewood and ‘‘0’’ for species not used
for those provisioning services.
We used Akaike Information Criterion (AIC) for model develop-
ment and selection and Akaike weights to determine optimal mod-
els (Burnham and Anderson, 2002). We calculated odds ratios from
parameter estimates to determine the influence of model factors
on predicted likelihood of a species being used as firewood or con-
struction. Odds ratios were calculated only for the factors in the
top models. A logit link function was used to model the logistic
regression. All statistical analyses were carried out using the statis-
tical program R (R Development Core Team, 2012). Petchey and
Gaston’s FD and the regression tree were calculated using R-code
adapted from Flynn et al. (2009). The optimal logistic regression
models were selected using MuMIn package (Barton, 2010).
3. Results
Two primary patterns were evident in our results: (i) there was
evidence of decreasing functional diversity (FD), utilitarian diver-
sity and species richness across all modified habitats and (ii) the
suite of traits most likely to be lost from forests was closely asso-
ciated with use of provisioning services. There were significant dif-
ferences in FD across the forest–agriculture matrix gradient
(Kruskal–Wallis, P < 0.001), with lowest FD in the most modified
environments (matrix) and highest in the least modified (forest
interior; post hoc pair-wise comparisons using Mann–Whitney,
P < 0.05; Fig. 2A). This pattern was not always consistent, however,
with interior–edge habitats showing statistically similar FD values
as matrix and interior–edge-matrix showing statistically similar
values as interior habitats (Fig. 2A). There were also significant dif-
ferences in utilitarian diversity across the forest–agriculture matrix
gradient (Kruskal–Wallis, P < 0.001; Fig. 2B), where the most mod-
ified environments had the lowest utilitarian diversity (post hoc
Mann–Whitney, P < 0.05; Fig. 2B). The same pattern was true for
species richness (Kruskal–Wallis, P < 0.001; Fig. 2C). FD, utilitarian
diversity and species richness also all declined significantly in con-
tinuous vs. fragmented forests (Mann–Whitney, P < 0.001 for all;
Fig. 3A–C).
The regression trees showed that species that were present in
the forest interior and absent from other categories were associ-
ated predominantly with bark thickness, WSG, height and second-
arily by SLA (R = 0.64, P < 0.001; Fig. 4). Thin-barked species, with
high WSG, and of relatively large stature and species with rela-
tively intermediate SLA values (P324 mm2/g) were more likely
to be absent from highly modified forests (Fig. 4). Species present
in fragmented sites were predominantly linked by large leaves
(P32.3 mm2), high leaf water content, and low-to-intermediate
SLA values (P263 mm2/g; Fig. 4).
The logistic regression was used to determine the association
between provisioning services (construction and firewood) and
functional traits. The Hosmer-Lemeshow statistic was not signifi-
cant, indicating there was no evidence for lack of model fit
(P = 0.703; data not shown). Additionally, the full model gives a
significantly better fit than the reduced model (P < 0.05;
X2 = 12.11, df = 6; log-likelihood = 41.57, df = 7). Bark thickness,
height and WSG were the best predictors of whether a plant was
more likely to be used for construction (Table 1). Additionally, bark
thickness was consistently an important predictor in all four top
models. The odds ratio showed that for every one unit increase
in bark thickness and WSG, the odds of species being used for con-
struction decreased by 15.1% and 97.5%, respectively (Table 1 and
Fig. 5A and C). Conversely, for every one unit increase in height,
the odds of species being used for construction increased by 12%
(Table 1 and Fig. 5B). The log ratio test indicated that the top can-
didate model for construction was not significant (P = 0.32;
Table 1).
The log ratio test indicated that the top candidate model for
firewood was significant (P < 0.05; Table 1). The top model in-
cluded three predictors: bark thickness, SLA and WSG (Table 1).
Leaf size, leaf water content and height were included in the other
top models, with bark thickness consistently shown as an impor-
tant predictor. The odds ratio showed that for every one unit in-
crease in bark thickness, there was a 22% increase in probability
of a species being used for firewood (Table 1; Fig. 5D). The odds ra-
tio for SLA showed that for every one unit increase in SLA, there
was a 0.6% decrease in the odds of species being used for firewood
(Table 1; Fig. 5E). For every one unit increase in WSG, the odds of
species being used for firewood increased by a factor of 133
(Table 1 and Fig. 5F).
4. Discussion
4.1. Changes in functional trait diversity, utilitarian diversity and
species richness
Regardless of whether land use change was measured on a for-
est-agricultural matrix gradient or by comparing continuous and
fragmented forests, species richness covaried with FD and utilitar-
ian diversity. Our results differ from that of Flynn et al. (2009), who
did not detect a decrease in plant functional diversity (also mea-
sured with Petchey and Gaston’s FD) as land use intensification in-
creased. This difference may be a consequence of our choice of
traits and focus on utilitarian plants, as well as differences in the
range of land use intensification categories. Similarly, Laliberté
et al. (2010) found that while the majority of plant groups re-
sponded negatively to land use intensification, certain groups in
tropical rainforests showed no net change as a result of intensifica-
tion. Conversely, Katovai et al. (2012) found similar patterns to our
study, showing that the most disturbed human-maintained land
uses in wet tropical forests had lower functional diversity than
remnant primary and secondary forests. This underscores the
rather nuanced effects of land use on plant functional traits, invit-
ing caution when making generalizations about trait responses,
especially in tropical systems. It is notable that our results are spe-
cific to traits associated with utilitarian species, which is a subset
of the species-pool in these forests. From a conservation perspec-
tive, it was necessary to focus on a suite of traits and species that
reflected anthropogenic influences on the ecosystem, and also,
one of the primary drivers of forest degradation in Madagascar is
harvesting of timber and non-timber forest products. Whether
 K.A. Brown et al. / Biological Conservation 161 (2013) 118–127 123

Fig. 2. Results from Kruskal–Wallis test of Petchey and Gaston’s FD (A); utilitarian diversity (B); and species richness (C) for forest–agriculture matrix gradient going from low
(Interior) to increasingly modified forests. There were significant differences in all three variables across the disturbance gradient (P < 0.001, Kruskal–Wallis). Significant pair-
wise differences between groups using post hoc Mann–Whitney test are indicated with different letters (P < 0.05).

the observed patterns remain after inclusion of non-utilitarian
plants will be addressed in future studies.
4.2. Functional trait loss in modified habitats and trait-provisioning
service relationship
The results from the CART and logistic regression were consis-
tent, in that the traits shown to be absent from modified habitats
were associated with species used for the two most important pro-
visioning services in the region: construction and firewood. Our re-
sults suggest a direct influence of subsistence communities in
Madagascar on plant assemblages, with people apparently choos-
ing species-specific traits for particular provisioning services. For
instance, plants with thin barks and relatively large-stature were
more likely to be lost (presumably removed) from the agricul-
ture-forest matrix gradient. We also identified the provisioning
service for which these traits were best associated, showing that
species with thin barks and large-stature were more likely to be
used for construction. The traits most associated with species used
for firewood were bark thickness, high WSG and SLA. The impor-
tance of the former two traits’ association with firewood as a pro-
visioning service make intuitive sense, since species with those
traits would provide a hotter, longer lasting burn (Vines, 1968).
For instance, species with thick barks are better adapted to survive
in a regime of frequent fires (Vines, 1968), but it is possibly that
trait property which makes such plants preferred for use as fire-
wood. Additionally, species with dense wood correlate with resis-
tance to drought-induced mortality (Preston et al., 2006),
mechanical breakage (Niklas, 1992), occur in later successional
stages (Falster and Westoby, 2005) and have lower mortality in
tropical forests (Kraft et al., 2010), properties that may help explain
the preference for such species for construction and firewood use.
Species with low-to-intermediate SLA values were more likely to
be used for firewood and more likely to be lost from modified envi-
ronments. SLA is positively related to relative growth rate and pho-
tosynthetic capacities (Wright, 2002) and negatively associated
124 K.A. Brown et al. / Biological Conservation 161 (2013) 118–127

Fig. 3. Results from Mann–Whitney test for Petchey and Gaston’s FD (A) utilitarian diversity (B) and species richness (C) for continuous vs. fragmented forests. There were
significant differences in all three variables across the disturbance gradient (P < 0.01). Asterisks indicate significance.

with solar irradiance (Long et al., 2011) and wood density (King
et al., 2006). Therefore, species with low-intermediate SLA were
more likely to have high wood density, indicating a trade-off be-
tween SLA and WSG, which suggests SLA may simply be associated
with an important trait on which people focus. It should be noted
that local residents are most likely not targeting traits we have
measured, at least not to the extent that they are choosing species
preferred for particular provisioning services. Decrease or loss of
certain traits is probably an inadvertent consequence of local resi-
dents relying on a subset of species for construction and firewood.
Nonetheless, these activities will influence community assem-
blages, particularly for regenerating forests.
There are other processes that do not implicate human activity
as a factor that affects changes in trait structure along forest–
agriculture matrix gradients. Coexistence theory suggests that
environmental filters, niche differentiation or neutral processes
can shape trait structure within plant communities. However, it
is difficult to differentiate between trait selection mediated by
anthropogenic activities from that regulated by natural processes.
For instance, the trait structure of secondary forests recovering
from or subjected to resource extraction may exhibit a similar
pattern as those modulated by niche differentiation: co-occurring
species within forests modulated by both may exhibit trait diver-
gence, leading to relatively higher variability in trait values (Kraft
et al., 2008; Paine et al., 2011). Neutral (niche-free) processes,
wherein dispersal and stochastic demographic processes predomi-
nate (Hubbell, 2001), may also lead to trait distributions with high
variability (Weiher et al., 2011). In complex social–ecological sys-
tems (Dawson et al., 2010) such as the forests presently under
study, trait selection will most likely be influenced by a combina-
tion of natural processes as well as direct and indirect human
activity, highlighting the importance of explicitly considering
how anthropogenic disturbances modulate trait structure. None-
theless, processes that govern species coexistence are quite varied
(Weiher et al., 2011; Weiher and Keddy, 1995) and our analyses
did not distinguish whether traits for co-occurring species in dis-
turbed assemblages were controlled by niche-based, niche-free
or a combination of these processes.
The implications that decrease in FD and utilitarian diversity
will have for local residents in this region are unclear. The likely
consequence might be a negative impact on people’s livelihood,
making it increasingly difficult to locate species that are preferred
for basic provisioning services. The degree of impact may depend
on local activities, as drivers of deforestation and forest degrada-
tion are often contextually specific (Ingram and Dawson, 2006);
this underscores the importance of using management policies tai-
lored for local conditions and addressing the causes and impacts of
these processes at a landscape scale. This approach may help
 K.A. Brown et al. / Biological Conservation 161 (2013) 118–127 125

Fig. 4. Regression tree showing traits present in forest interior (least modified habitat along forest–agriculture matrix gradient) and absent from all other categories; and
regression tree showing traits present in continuous and absent from fragmented forests. Species are sorted by traits into groups at the tips of the regression trees. The value
at each tip indicates the proportion of species that were present in at least one plot in the interior of the forest but were absent from the matrix; or proportion of species
present in at least one plot in the continuous forest but absent from fragments.

Table 1
Top candidate models for the multiple logistic regression of species used for construction or firewood as a function of bark thickness, height, leaf size, leaf water content, SLA and
WSG. Information is given for the top four models, along with their associated Akaike Information criterion (AIC), DAIC, Akaike weights (wi), and the parameters included in the
model. Models are ranked by wi (highest value). Bold indicates model significance at P < 0.05).

Model Residual deviance AIC DAIC wi Parameters

Construction
1 78.9 87.5 0.00 0.118 Bark + Height + WSG
2 81.3 87.6 0.13 0.110 Bark + WSG
3 84.2 88.4 0.87 0.076 Bark
4 78.1 89.0 1.51 0.055 Bark + Height + Leaf size + WSG
Firewood
1 84.4 92.9 0.00 0.250 Bark + SLA + WSG
2 83.3 94.2 1.29 0.131 Bark + Leaf size + SLA + WSG
3 83.8 94.7 1.80 0.101 Bark + Leaf water + SLA + WSG
4 84.2 95.1 2.20 0.083 Bark + Height + SLA + WSG

Odds ratio: BarkConstruction = 0.849, HeightConstruction = 1.12, WSGConstruction = 0.025; BarkFirewood = 1.22, SLAFirewood = 0.994, WSGFirewood = 133.0.
Significance tests for the best models were conducted using model likelihood ratio tests. The X2 statistic (df, n) and significance are shown: X 2Construction (6, 75) = 7.01, P = 0.32;
X 2Firewood (6, 75) = 12.11, P = 0.05.

development of a conceptual framework that allows for prediction
of how disturbances will affect tropical forest biodiversity, using
knowledge of local residents’ practice and preference for provision-
ing services to inform such a framework. Conservation efforts
should recognize that the factors that promote deforestation and
forest degradation are a consequence of people’s efforts to main-
tain their basic livelihood, and require local-scale interventions
that feed into landscape-scale policy initiatives. This can take the
form of a participatory adaptive management approach integrated
within a framework that allows adapting to the needs of local
stakeholders, yet also responds to the highly dynamic nature of
these forest landscapes.
126 K.A. Brown et al. / Biological Conservation 161 (2013) 118–127
Fig. 5. Predicted probability plots for best candidate models for the logistic regression of species used for construction or firewood as a function of bark thickness, height, leaf
size, leaf water content, SLA and WSG. The best three predictors for construction and firewood are shown. Top row shows construction: (A) Bark thickness (mm), (B) Height
(cm), (C) WSG. Bottom row shows firewood: (D) Bark thickness (mm), (E) SLA (mm2/g), and (F) WSG.
As an example of such an adaptive management initiative, a for-
est restoration project should accommodate the social–ecological
complexities of the region. The plants chosen for reforestation
could be determined both from interviews with local village ex-
perts who identify provisioning services vital for their livelihood
as well as researchers who attempt to identify plant traits that
are necessary to restore certain ecosystem functions. Both would
need to continuously adapt, the former to local residents switching
from one preferred resource to another; the latter responding to
new research that informs trait—ecosystem function relationships.
Such an approach addresses the social–ecological dynamics of
tropical landscapes, while also responding to both the needs of lo-
cal people and the loss of biodiversity. Whichever approach is
adopted, it must recognize the resources required for people’s live-
lihoods and seek to maintain functioning ecosystems on which
they depend.
In conclusion, our study shows that as forests in southeastern
Madagascar are converted from mature to secondary forests to
agro-forestry matrix, they are losing a specific, non-random suite
of traits. The association between trait loss and provisioning ser-
vices suggest that humans directly influence plant traits in humid
forests of Madagascar. Moreover, our results suggest that targeted
harvesting for specific plant characteristics likely has produced the
resulting overall trait composition in these assemblages. Rural peo-
ple, living on the forest fringe influence the distribution of traits in
plant assemblages, which in turn will not only affect the structure
of regenerating forests, but will also influence future community
assembly and trait expression in these forests.
Acknowledgements
We thank Madagascar National Parks (MNP), Madagascar Insti-
tute pour la Conservation des Environnements Tropicaux (MICET),
Institute for the Conservation of Tropical Environments (ICTE),
and Madagascar Biodiversity Partnership (MBP) for valuable logical
support. We thank Dan F. B. Flynn for comments on earlier versions
of the manuscript. We thank two anonymous reviewers for
comments on the manuscript. Funding was provided by Kingston
University Enterprise Fund (KAB), David and Lucile Packard Foun-
dation (PCW), SUNY-Stony Brook (PCW), University of Calgary
(SEJ, KED, SMH), Primate Conservation Inc. (SEJ, KED, SMH), Con-
servation International and Margo Marsh Biodiversity Foundation
Primate Action Fund (SEJ, KED, SMH), American Society of Prima-
tologists (KED), International Primatological Society (SMH), and
Calgary Zoological Society (SMH).
References
Abram, N., 2008. People and Protected Area Conflict: unravelling the Complexities
of Illegal Resource Use by Forest Dependent Communities in Manombo Forest,
Madagascar. Imperial College London, London.
Achard, F., Eva, H.D., Mayaux, P., Stibig, H.J., Belward, A., 2004. Improved estimates
of net carbon emissions from land cover change in the tropics for the 1990s.
Global Biogeochem. Cycles 18.
Agarwal, D.K., Silander, J.A., Gelfand, A.E., Dewar, R.E., Mickelson, J.G., 2005. Tropical
deforestation in Madagascar: analysis using hierarchical spatially explicit,
Bayesian regression models. Ecol. Model. 185, 105–131.
Asner, G.P., Rudel, T.K., Aide, T.M., Defries, R., Emerson, R., 2009. A contemporary
assessment of change in humid tropical forests. Conserv. Biol. 23, 1386–1395.
Barton, K., 2010. Multi-model Inference (MuMIn). R Statistical Package 2.14.2.
Brown, K.A., Ingram, J.C., Flynn, D.F.B., Razafindrazaka, R., Jeannoda, V., 2009.
Protected areas safeguard tree and shrub communities from degradation and
invasion: a case study in Eastern Madagascar. Environ. Manage. 44, 136–148.
Brown, K.A., Flynn, D.F.B., Abram, N.K., Ingram, J.C., Johnson, S.E., Wright, P., 2011.
Assessing natural resource use by forest-reliant communities in madagascar
using functional diversity and functional redundancy metrics. Plos One 6.
Burnham, K.P., Anderson, D.R., 2002. Model selection and multimodel inference: a
practical information-theoretic approach, second ed. Springer-Verlag, New
York.
Cardinale, B.J., Duffy, J.E., Gonzalez, A., Hooper, D.U., Perrings, C., Venail, P., Narwani,
A., Mace, G.M., Tilman, D., Wardle, D.A., Kinzig, A.P., Daily, G.C., Loreau, M.,
Grace, J.B., Larigauderie, A., Srivastava, D.S., Naeem, S., 2012. Biodiversity loss
and its impact on humanity. Nature 486, 59–67.
Consiglio, T., Schatz, G.E., McPherson, G., Lowry, P.P., Rabenantoandro, J., Rogers,
Z.S., Rabevohitra, R., Rabehevitra, D., 2006. Deforestation and plant diversity of
Madagascar’s littoral forests. Conserv. Biol. 20, 1799–1803.
Cornelissen, J.H.C., Lavorel, S., Garnier, E., Diaz, S., Buchmann, N., Gurvich, D.E.,
Reich, P.B., ter Steege, H., Morgan, H.D., van der Heijden, M.G.A., Pausas, J.G.,
Poorter, H., 2003. A handbook of protocols for standardised and easy
measurement of plant functional traits worldwide. Aust. J. Bot. 51, 335–380.
 K.A. Brown et al. / Biological Conservation 161 (2013) 118–127
Cornwell, W.K., Cornelissen, J.H.C., Amatangelo, K., Dorrepaal, E., Eviner, V.T., Godoy,
O., Hobbie, S.E., Hoorens, B., Kurokawa, H., Perez-Harguindeguy, N., Quested,
H.M., Santiago, L.S., Wardle, D.A., Wright, I.J., Aerts, R., Allison, S.D., van
Bodegom, P., Brovkin, V., Chatain, A., Callaghan, T.V., Diaz, S., Garnier, E.,
Gurvich, D.E., Kazakou, E., Klein, J.A., Read, J., Reich, P.B., Soudzilovskaia, N.A.,
Vaieretti, M.V., Westoby, M., 2008. Plant species traits are the predominant
control on litter decomposition rates within biomes worldwide. Ecol. Lett. 11,
1065–1071.
Dawson, T.P., Rounsevell, M.D.A., Kluvankova-Oravska, T., Chobotova, V., Stirling, A.,
2010. Dynamic properties of complex adaptive ecosystems: implications for the
sustainability of service provision. Biodivers. Conserv. 19, 2843–2853.
de Bello, F., Lavorel, S., Diaz, S., Harrington, R., Cornelissen, J.H.C., Bardgett, R.D.,
Berg, M.P., Cipriotti, P., Feld, C.K., Hering, D., da Silva, P.M., Potts, S.G., Sandin, L.,
Sousa, J.P., Storkey, J., Wardle, D.A., Harrison, P.A., 2010. Towards an assessment
of multiple ecosystem processes and services via functional traits. Biodivers.
Conserv. 19, 2873–2893.
De’ath, G., Fabricius, K.E., 2000. Classification and regression trees: a powerful yet
simple technique for ecological data analysis. Ecology 81, 3178–3192.
Delmore, K.E., 2009. Maintenance of stability in the Andringitra Brown Lemur
hybrid zone. In: Department of Anthropology. University of Calgary, Calgary.
Faith, D.P., 1992. Conservation evaluation and phylogenetic diversity. Biol. Conserv.
61, 1–10.
Falster, D.S., Westoby, M., 2005. Alternative height strategies among 45 dicot rain
forest species from tropical Queensland, Australia. J. Ecol. 93, 521–535.
Flynn, D.F.B., Gogol-Prokurat, M., Nogeire, T., Molinari, N., Richers, B.T., Lin, B.B.,
Simpson, N., Mayfield, M.M., DeClerck, F., 2009. Loss of functional diversity
under land use intensification across multiple taxa. Ecol. Lett. 12, 22–33.
Flynn, D.F.B., Mirotchnick, N., Jain, M., Palmer, M.I., Naeem, S., 2011. Functional and
phylogenetic diversity as predictors of biodiversity–ecosystem-function
relationships. Ecology 92, 1573–1581.
Foley, J.A., Asner, G.P., Costa, M.H., Coe, M.T., DeFries, R., Gibbs, H.K., Howard, E.A.,
Olson, S., Patz, J., Ramankutty, N., Snyder, P., 2007. Amazonia revealed: forest
degradation and loss of ecosystem goods and services in the Amazon Basin.
Front. Ecol. Environ. 5, 25–32.
Gardner, T.A., Barlow, J., Chazdon, R., Ewers, R.M., Harvey, C.A., Peres, C.A., Sodhi,
N.S., 2009. Prospects for tropical forest biodiversity in a human-modified world.
Ecol. Lett. 12, 561–582.
Green, G.M., Sussman, R.W., 1990. Deforestation history of the eastern rain forests
of Madagascar from satellite images. Science 248, 212–215.
Harper, G.J., Steininger, M.K., Tucker, C.J., Juhn, D., Hawkins, F., 2007. Fifty years of
deforestation and forest fragmentation in Madagascar. Environ. Conserv. 34,
325–333.
Holmes, S.M., 2012. Habitat use and population genetics of the Black-and-White
Ruffed lemur (Varecia variegata) in a fragmented landscape in southeastern
Madagascar. In: Department of Anthropology. University of Calgary, Calgary.
Hubbell, S.P., 2001. The Unified Neutral Theory of Biodiversity and Biogeography.
Princeton University Press, Princeton, New Jersey, USA.
Ingraldi, C., 2010. Forest fragmentation and edge effects of eight sympatric lemur
species in southeast Madagascar. In: Department of Anthropology. University of
Calgary, Calgary.
Ingram, J.C., Dawson, T.P., 2006. Forest cover, condition, and ecology in human-
impacted forests, south-eastern Madagascar. Conserv. Soc. 4, 194–230.
Irwin, M.T., Wright, P.C., Birkinshaw, C., Fisher, B.L., Gardner, C.J., Glos, J., Goodman,
S.M., Loiselle, P., Robeson, P., Raharison, J.L., Raherilalao, M.J., Rakotondravony,
D., Raselimanana, A., Ratsimbazafy, J., Sparks, J.S., Wilme, L., Ganzhorn, J.U.,
2010. Patterns of species change in anthropogenically disturbed forests of
Madagascar. Biol. Conserv. 143, 2351–2362.
Johnson, S.E., 2002. Ecology and Speciation in Brown lemurs: White-Collared
lemurs (Eulemur albocollaris) and hybrids (Eulemur albocollaris  Eulemur
fulvus rufus) in southeastern Madagascar. University of Texas at Austin, Austin.
Katovai, E., Burley, A.L., Mayfield, M.M., 2012. Understory plant species and
functional diversity in the degraded wet tropical forests of Kolombangara
Island, Solomon Islands. Biol. Conserv. 145, 214–224.
Kessler, M., Abrahamczyk, S., Bos, M., Buchori, D., Putra, D.D., Gradstein, S.R., Hohn,
P., Kluge, J., Orend, F., Pitopang, R., Saleh, S., Schulze, C.H., Sporn, S.G., Steffan-
Dewenter, I., Tjitrosoedirdjo, S., Tscharntke, T., 2009. Alpha and beta diversity of
plants and animals along a tropical land-use gradient. Ecol. Appl. 19, 2142–
2156.
King, D.A., Davies, S.J., Tan, S., Noor, N.S.M., 2006. The role of wood density and stem
support costs in the growth and mortality of tropical trees. J. Ecol. 94, 670–680.
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.
Kraft, N.J.B., Metz, M.R., Condit, R.S., Chave, J., 2010. The relationship between wood
density and mortality in a global tropical forest data set. New Phytol. 188,
1124–1136.
Kremen, C., 2005. Managing ecosystem services: what do we need to know about
their ecology? Ecol. Lett. 8, 468–479.
Kremen, C., Raymond, I., Lance, K., 1998. An interdisciplinary tool for monitoring
conservation impacts in Madagascar. Conserv. Biol. 12, 549–563.
Laliberté, E., Wells, J.A., DeClerck, F., Metcalfe, D.J., Catterall, C.P., Queiroz, C., Aubin,
I., Bonser, S.P., Ding, Y., Fraterrigo, J.M., McNamara, S., Morgan, J.W., Merlos, D.S.,
127
Vesk, Mayfield, M.M., 2010. Land-use intensification reduces functional
redundancy and response diversity in plant communities. Ecol. Lett. 13, 76–86.
Laurance, W.F., Nascimento, H.E.M., Laurance, S.G., Andrade, A., Ewers, R.M., Harms,
K.E., Luizao, R.C.C., Ribeiro, J.E., 2007. Habitat fragmentation, variable edge
effects, and the landscape-divergence hypothesis. Plos One 2.
Lavorel, S., Grigulis, K., Lamarque, P., Colace, M.P., Garden, D., Girel, J., Pellet, G.,
Douzet, R., 2011. Using plant functional traits to understand the landscape
distribution of multiple ecosystem services. J. Ecol. 99, 135–147.
Lin, B.B., Flynn, D.F.B., Bunker, D.E., Uriarte, M., Naeem, S., 2011. The effect of
agricultural diversity and crop choice on functional capacity change in
grassland conversions. J. Appl. Ecol. 48, 609–618.
Long, W.X., Zang, R.G., Schamp, B.S., Ding, Y., 2011. Within- and among-species
variation in specific leaf area drive community assembly in a tropical cloud
forest. Oecologia 167, 1103–1113.
Luck, G.W., Chan, K.M.A., Fay, J.P., 2009. Protecting ecosystem services and
biodiversity in the world’s watersheds. Conserv. Lett. 2, 179–188.
Mayfield, M.M., Boni, M.E., Daily, G.C., Ackerly, D., 2005. Species and functional
diversity of native and human-dominated plant communities. Ecology 86,
2365–2372.
Mayfield, M.M., Bonser, S.P., Morgan, J.W., Aubin, I., McNamara, S., Vesk, P.A., 2010.
What does species richness tell us about functional trait diversity? Predictions
and evidence for responses of species and functional trait diversity to land-use
change. Glob. Ecol. Biogeogr. 19, 423–431.
McGill, B.M., Sutton-Grier, A.E., Wright, J.P., 2010. Plant trait diversity buffers
variability in denitrification potential over changes in season and soil
conditions. Plos One 5.
Mittermeier, R.A., Gil, P.R., Hoffman, M., Pilgrim, J., Brooks, T., Mittermeier, C.G.,
Lamoreus, J., da Fonseca, G.A.B., 2005. Hotspots revisited: Earth’s biologically
richest and most endangered terrestrial ecoregions. The University of Chicago
Press, Washington, DC, USA.
Morris, R.J., 2010. Anthropogenic impacts on tropical forest biodiversity: a network
structure and ecosystem functioning perspective. Philos. Trans. R. Soc. B – Biol.
Sci. 365, 3709–3718.
Mouillot, D., Villeger, S., Scherer-Lorenzen, M., Mason, N.W.H., 2011. Functional
structure of biological communities predicts ecosystem multifunctionality. Plos
One 6.
Muller-Landau, H.C., 2004. Interspecific and inter-site variation in wood specific
gravity of tropical trees. Biotropica 36, 20–32.
Myers, N., Mittermeier, R.A., Mittermeier, C.G., da Fonseca, G.A.B., Kent, J., 2000.
Biodiversity hotspots for conservation priorities. Nature 403, 853–858.
Naeem, S., Duffy, J.E., Zavaleta, E., 2012. The functions of biological diversity in an
age of extinction. Science 336, 1401–1406.
Niklas, K., 1992. Plant Biomechanics. University of Chicago Press, Chicago, USA.
Norris, K., Asase, A., Collen, B., Gockowksi, J., Mason, J., Phalan, B., Wade, A., 2010.
Biodiversity in a forest–agriculture mosaic – the changing face of West African
rainforests. Biol. Conserv. 143, 2341–2350.
Paine, C.E.T., Baraloto, C., Chave, J., Herault, B., 2011. Functional traits of individual
trees reveal ecological constraints on community assembly in tropical rain
forests. Oikos 120, 720–727.
Pardini, R., Faria, D., Accacio, G.M., Laps, R.R., Mariano-Neto, E., Paciencia, M.L.B.,
Dixo, M., Baumgarten, J., 2009. The challenge of maintaining Atlantic forest
biodiversity: a multi-taxa conservation assessment of specialist and generalist
species in an agro-forestry mosaic in southern Bahia. Biol. Conserv. 142, 1178–
1190.
Preston, K.A., Cornwell, W.K., DeNoyer, J.L., 2006. Wood density and vessel traits as
distinct correlates of ecological strategy in 51 California coast range
angiosperms. New Phytol. 170, 807–818.
Quetier, F., Thebault, A., Lavorel, S., 2007. Plant traits in a state and transition
framework as markers of ecosystem response to land-use change. Ecol. Monogr.
77, 33–52.
R_Development_Core_Team, 2012. R: A language and environment for statistical
computing. In: R Foundation for Statistical Computing. Vienna, Austria.
Srivastava, D.S., Vellend, M., 2005. Biodiversity–Ecosystem function research: is it
relevant to conservation? Annu. Rev. Ecol. Evol. Syst. 36, 267–294.
Vines, R.G., 1968. Heat transfer through bark, and the resistance of trees to fire.
Aust. J. Bot. 16, 499–514.
Violle, C., Navas, M.L., Vile, D., Kazakou, E., Fortunel, C., Hummel, I., Garnier, E., 2007.
Let the concept of trait be functional! Oikos 116, 882–892.
Weiher, E., Keddy, P.A., 1995. Assembly rules, null models, and trait dispersion: new
questions front old patterns. Oikos 74, 159–164.
Weiher, E., Freund, D., Bunton, T., Stefanski, A., Lee, T., Bentivenga, S., 2011.
Advances, challenges and a developing synthesis of ecological community
assembly theory. Philos. Trans. R. Soc. B – Biol. Sci. 366, 2403–2413.
Whitmore, T.C., 1998. An Introduction to Tropical Rain Forests. Oxford University
press, Oxford, UK.
Williamson, G.B., Wiemann, M.C., 2010. Measuring wood specific gravity –
correctly. Am. J. Bot. 97, 519–524.
Wright, S.J., 2002. Plant diversity in tropical forests: a review of mechanisms of
species coexistence. Oecologia 130, 1–14.