Linking patterns and processes of tree community assembly across
spatial scales in tropical montane forests
MANUEL J. MAC
NIGO GRANZOW-DE LA CERDA,
MARTINS DE CARVALHO,
CARLOS I. ESPINOSA,
NATHAN G. SWENSON,
AND LUIS CAYUELA
Departamento de Biolog
ısica y Qu
anica, Universidad Rey Juan Carlos, Calle Tulip
an s/n, M
ES- 28933 Spain
Departamento de Biolog
Area de Bot
anica, Universidad Aut
onoma de Madrid, Calle Darwin 2, Madrid ES-28049 Spain
Centro de Investigaci
on en Biodiversidad y Cambio Global (CIBC-UAM), Universidad Aut
onoma de Madrid, Calle Darwin 2,
Madrid ES-28049 Spain
Herbario HUTPL, Departamento de Ciencias Biol
ogicas, Universidad T
ecnica Particular de Loja, A. P. 11-01-608, Loja, Ecuador
Departamento de Ciencias Biol
ogicas, Universidad T
ecnica Particular de Loja, A.P. 11-01-608, Loja, Ecuador
ımica, Pontificia Universidad Cat
olica del Per
u, A.P. Lima 32, Lima, Per
ECI, School of Geography and Environment, University of Oxford, OX1 3QY Oxfordshire, UK
Department of Biology, University of Maryland, College Park, Maryland 20742 USA
nares-de-Dios, G., M. J. Mac
I. Granzow-de la Cerda, I. Arnelas, G. Martins de Carvalho,
C. I. Espinosa, N. Salinas, N. G. Swenson, and L. Cayuela. 2020. Linking patterns and processes of tree
community assembly across spatial scales in tropical montane forests. Ecology 101(7):e03058. 10.1002/ecy.
Abstract. Many studies have tried to assess the role of both deterministic and stochastic
processes in community assembly, yet a lack of consensus exists on which processes are more
prevalent and at which spatial scales they operate. To shed light on this issue, we tested two
nonmutually exclusive, scale-dependent hypotheses: (1) that competitive exclusion dominates
at small spatial scales; and (2) that environmental filtering does so at larger ones. To accom-
plish this, we studied the functional patterns of tropical montane forest communities along two
altitudinal gradients, in Ecuador and Peru, using floristic and functional data from 60 plots of
0.1 ha. We found no evidence of either functional overdispersion or clustering at small spatial
scales, but we did find functional clustering at larger ones. The observed pattern of clustering,
consistent with an environmental filtering process, was more evident when maximizing the
environmental differences among any pair of plots. To strengthen the link between the
observed community functional pattern and the underlying process of environmental filtering,
we explored differences in the climatic preferences of the most abundant species found at lower
and higher elevations and examined whether their abundances shifted along the elevation gra-
dient. We found (1) that greater community functional differences (observed between lower
and upper tropical montane forest assemblies) were mostly the result of strong climatic prefer-
ences, maintained across the Neotropics; and (2) that the abundances of such species shifted
along the elevational gradient. Our findings support the conclusion that, at large spatial scales,
environmental filtering is the overriding mechanism for community assembly, because the pat-
tern of functional clustering was linked to species’similarities in their climatic preferences,
which ultimately resulted in shifts in species abundances along the gradient. However, there
was no evidence of competitive exclusion at more homogeneous, smaller spatial scales, where
plant species effectively compete for resources.
Key words: altitudinal gradients; community assembly; competitive exclusion; environmental filtering;
functional traits; spatial scale; tropical montane forest.
Which processes determine the structure and composi-
tion of communities? This question has generated dis-
cussion among naturalists for decades (Clements 1916,
Gause 1934, MacArthur and Levins 1967, Wright 2002),
becoming almost an obsession for ecologists and the
core of community assembly research (Sutherland et al.
2013). In the quest for answers, scholars have proposed
different explanations for the mechanisms shaping the
structure and distribution of natural communities, with
environmental filtering and competitive exclusion being
among the most broadly embraced (Belyea and Lan-
caster 1999, G€
otzenberger et al. 2012, Kraft and Ackerly
2014). According to environmental filtering, the abiotic
milieu acts as a sieve, allowing only species with certain
traits or phenotypes to establish and survive successfully,
Manuscript received 29 May 2019; revised 4 November 2019;
accepted 5 December 2019. Corresponding Editor: Daniel
Article e03058; page 1
Ecology, 101(7), 2020, e03058
©2020 by the Ecological Society of America
whereas the rest fail (Bazzaz 1991, Weiher and Keddy
ıaz et al. 1998). However, competitive exclusion
posits that coexisting species compete for the resources
until one excludes the others. Thus, the less two species
compete for resources, the more likely they are to co-oc-
cur (Diamond 1975, Abrams 1996, Dayan and Sim-
berloff 2005). As an alternative to these niche-based
explanations, the neutral assembly theory considers that
all species are ecologically equivalent, and therefore
communities would be the result of only dispersal limita-
tion events and stochastic demographic processes (Hub-
To address the community assembly issue, functional
ecology has emerged as a more powerful and suitable tool
than classical taxonomic approaches based on Linnaean
binomials, because indices of species composition and
abundances provide little information about the ecologi-
cal strategies of those species (Fukami et al. 2005, Swen-
son 2012). Instead, the functional approach relies on
species functional traits—easily measurable morphologi-
cal or physiological characters of individuals relevant to
growth, survival, or reproduction (Westoby and Wright
2006, Funk et al. 2017)—as proxies of ecological perfor-
mance and, consequently, capable of explaining how spe-
cies interact with their abiotic and biotic environment
(Keddy 1992, McGill et al. 2006, Violle et al. 2007). By
using the functional ecology framework, environmental
filtering has been identified as a major force shaping com-
munities across a number of biomes, including drylands
(Le Bagousse-Pinguet et al. 2017), alpine (de Bello et al.
opez-Angulo et al. 2018), temperate (Cornwell
and Ackerly 2009) and tropical forests (Kraft et al. 2008,
Swenson and Enquist 2009, Lebrija-Trejos et al. 2010,
Baraloto et al. 2012). Nevertheless, the overwhelming
importance of environmental filtering on community
assembly has lately been questioned based on a disregard
towards spatial scale consideration (Chase 2014) and on
the uncertainty regarding whether functional patterns can
reliably indicate mechanisms (Mayfield and Levine 2010).
To account for spatial scale is of utmost importance in
ecology (Wiens 1989, Levin 1992, McGill 2010), and its
implications on community assembly are undeniable
(Whittaker et al. 2001, Kneitel and Chase 2004, Snyder
and Chesson 2004, M€
uller et al. 2013). For
instance, at a broad spatial scale, environmental filtering
seems to prevail over other processes (e.g., cacti do not
thrive in artic regions, nor do polar bears in the rain-
forest), and competitive exclusion has virtually no effect
on individuals that are many kilometers apart, although
it may have an effect at a smaller spatial scale, on indi-
viduals in close proximity. Nonetheless, ignoring spatial
scale-related implications has often led to discrepancies
about which processes dominate community assembly
(discussed by Chase 2014, Chalmandrier et al. 2017). To
incorporate the scale issue a hierarchical model has been
proposed, according to which assembly mechanisms
operate sequentially at different spatial scales (Weiher
and Keddy 1995b,G
otzenberger et al. 2012,
HilleRisLambers et al. 2012). This model encompasses
evolutionary and biogeographic processes, such as his-
torical patterns of speciation, extinction, or migration at
large/regional scales; to abiotic and biotic processes, like
environmental filtering or competitive exclusion at smal-
ler/local scales. The former processes define a regional
pool of potential colonizer species over which the latter
operate at finer scales, yielding the final assembly of
local communities. Under this paradigm, shifting the
scope of the studied community and the species pool will
allow us to clarify whether distinct assembly processes
are restricted to operate at certain spatial scales (Colwell
and Winkler 1984, Weiher and Keddy 1995b, Swenson
et al. 2007). For instance, environmental filtering may be
more prevalent when the species pool is defined from a
broad area encompassing strong abiotic heterogeneity,
for example, steep environmental or habitat gradients,
while the studied community, established at a relatively
narrower spatial scale, is constrained to an environmen-
tally homogeneous area (de Bello et al. 2013b, Garzon-
Lopez et al. 2014, reviewed in Kraft et al. 2015).
Community patterns based on co-occurring species
composition and abundance combined with functional
diversity have been broadly trusted to reflect different
community assembly processes. However, interpreting
them as unequivocal signals of actual assembly processes
is arguable. Traditionally and according to the commu-
nity assembly mechanisms above, two mutually exclud-
ing scenarios for species co-occurrence have been
proposed: (1) species could diverge in their ecological
strategies to achieve co-occurrence by avoiding competi-
tive exclusion, thus functional overdispersion should be
observed within the community (Watkins and Wilson
2003, Silvertown 2004, Stubbs and Wilson 2004, Caven-
der-Bares et al. 2009); or (2) species could converge in
their ecological strategies such as to enable them to
thrive in the same abiotic environment, resulting instead
in functional clustering (Keddy 1992, Cornwell et al.
2006, Ackerly and Cornwell 2007). In addition, if the
traits are phylogenetically conserved, a phylogenetic
overdispersion or clustering pattern should also be
observed within the community, respectively (Webb et al.
2002). This dichotomy, nevertheless, is an oversimplifica-
tion, because both theory (Chesson 2000, Grime 2006,
Mayfield and Levine 2010) and practice (Burns and
Strauss 2011, Narwani et al. 2013, Godoy et al. 2014)
have proven it to be unwarranted, since clustering can
also result from competitive exclusion as well as from
other biotic processes. For example, the existence of
selective herbivores (Uriarte 2000), pathogens (Parker
et al. 2015), or pollinators (Sargent and Ackerly 2008)
can specifically harm or favor groups of species exhibit-
ing certain traits, thus affecting the whole community
structure. Therefore, inferring unequivocally that envi-
ronmental filtering drives community assembly just on
the basis that a clustered functional pattern has been
observed could be deluding (G€
otzenberger et al. 2016,
Cadotte and Tucker 2017).
Article e03058; page 2 GUILLERMO BA
NARES-DE-DIOS ET AL. Ecology, Vol. 101, No. 7
Our study aims to advance the understanding of com-
munity assembly accounting for spatial scale implica-
tions and avoiding an overreliance on community
patterns. To achieve this, we studied functional diversity
of woody plant communities along two elevational gra-
dients in Andean tropical montane forests, one of the
most complex and diverse ecosystems worldwide. We
consider that at a given spatial scale only one single
mechanism will have a heavier influence on community
assembly. Under this premise we assume that, whereas
competitive exclusion would mainly operate at small
spatial scales at which the environment is relatively
homogeneous, environmental filtering would chiefly
emerge as a significant force shaping community assem-
bly at larger scales at which there is enough environmen-
tal heterogeneity to trigger a functional community
response. Therefore, we hypothesized that (1) if competi-
tive exclusion drives the community assembly at small
spatial scales (e.g., across neighboring individuals within
a plot), functional patterns would be overdispersed in
comparison to a null expectation and (2) if environmen-
tal filtering rule at larger scales (e.g., across plots spaced
hundreds of meters or at different elevations), functional
patterns would be clustered. Furthermore, following the
guidelines from Kraft et al. (2015) to state the impor-
tance of the processes in the assembly of the community
clearly, we sought evidences that link the observed com-
munity patterns with the underlying assembly processes
(e.g., correlations between thermal gradient and commu-
nity functional patterns or species abundances, respec-
tively). By doing so, this study will further contribute to
unmask the spatial scales at which different assembly
MATERIALS AND METHODS
Study regions and field sampling
The study was conducted along two elevational gradi-
ents of Andean tropical montane forest: one in Podocar-
pus National Park (Ecuador) and the other in Rio Abiseo
National Park (Peru) (Fig. 1). These sites were chosen
because both extend along wide elevational ranges (ca.
2,000 m), over a continuous forest cover, each within a
single river basin: the Bombuscaro River in Ecuador and
the Montecristo-Abiseo River in Peru. Three elevational
belts were defined at each site (lower, 800–1,100 m above
sea level [a.s.l.]; intermediate, 1,900–2,100 m a.s.l.; upper,
2,700–2,900 m a.s.l.). At each belt, 10 plots of 0.1 ha
(50 920 m) were established between 2015 and 2017 fol-
lowing Arellano et al. (2016) (Appendix S1: Table S1).
Each plot was subdivided in 10 subplots of 0.01 ha
(10 910 m). Plots were placed at least 300 m apart,
avoiding areas visibly affected by natural disturbances
(e.g., gaps caused by fallen trees or landslides). In each
plot, all woody individuals ≥2.5-cm diameter at breast
height (DBH) rooted within the plot limit were invento-
ried and their height estimated, although for this study
solely trees, treelets, shrubs, and lianas were taken into an
account. At least one voucher from every taxon was col-
lected for identification. Overall, 18,272 individuals were
inventoried in 60 plots: 9,366 in Ecuador and 8,905 in
Peru (Appendix S1: Table S1; Data S1).
Floristic data and functional characterization
Voucher specimens were identified at different her-
baria from Ecuador (HUTPL, LOJA, QCA) and Peru
(HAO, HUT, MOL, USM), acronyms according to
Thiers (n.d.. After thorough taxonomic effort, 424 indi-
viduals from Ecuador (4.5%) and 733 from Peru (8.23%)
could not be reliably assigned to a morphospecies level
and were therefore excluded from the analysis
(Appendix S1: Table S2). For Ecuador, the 8,942 identi-
fiable individuals were assigned to 734 taxa, comprising
471 species and 263 morphospecies. For Peru, the 8,172
identifiable individuals were assigned to 526 taxa,
including 189 species and 337 morphospecies. Standard-
ization of taxonomic species names was conducted using
the R package ‘Taxonstand’(Cayuela et al. 2012, 2017).
For each taxon, the following functional traits were
measured: specific leaf area (SLA), leaf thickness (LT),
and wood density (WD). These traits address key woody
plant functional strategy axes on which assembly mecha-
nisms operate (Wright et al. 2004, Kraft et al. 2008, Bar-
aloto et al. 2010). SLA was calculated from five leaves as
the ratio of leaf surface area (measured with a portable
laser leaf area meter CI- 202, CID Bio-Science, Camas,
Washington, USA) to leaf dry mass (after drying at 80°C
for 48 h). LT was measured with a digital calliper. Branch
wood density was used as a proxy for WD, as both are
strongly and positively correlated (Swenson and Enquist
2008). Sections of branches ca. 10 cm in length, as cylin-
drical as possible, were stripped of bark, and their diame-
ter and length measured in the field with a calliper to
determine their fresh volume. Density of the branch sec-
tion was calculated by dividing its fresh volume by its dry
mass (after drying at 80°Cfor48–72 h). Mean trait values
were calculated for every taxon (Data S1). All these pro-
tocols were based on Cornelissen et al. (2003) with just
one exception: for SLA and LT, leaves in full sun, at the
upper canopy, were avoided in order to make these traits
comparable between canopy and understory species.
Functional data were collected for 723 taxa in Ecuador
(98.5% of the total identified), that include 8,903 individ-
uals (95% of the total inventoried) and for 504 taxa in
Peru (95.8% of the total identified), that include 8,016
individuals (90.01% of the total inventoried). For a sum-
mary of community functional characterization, see
Appendix S1: Table S3.
Community assembly functional patterns
To elucidate how the consideration of different spatial
scales may influence the relative effect of distinct deter-
ministic mechanisms, the observed and the null
July 2020 COMMUNITY ASSEMBLY: PATTERNS AND PROCESSES Article e03058; page 3
community functional trait distribution patterns were cal-
culated for two spatial scale–related hypotheses (Fig. 2):
(1) trait distribution within a subplot compared with trait
distribution among nonadjacent subplots from the same
plot (small spatial scales), and (2) trait distribution within
a plot compared with trait distribution among plots
located along the elevational gradient (large spatial
scales). Deviations in the observed distribution pattern
from the null expectation would suggest the existence of
different deterministic community assembly processes,
such as environmental filtering or competitive exclusion,
whereas a close match between distributions could be
interpreted as evidence of stochastic community assembly
(Connor and Simberloff 1979, Gotelli and Graves 1996).
For both cases, the observed community trait distribu-
tion was calculated as U
are the trait Euclidean distances between pairs of
co-occurring individuals (for all the individuals within
each site) from distinct taxa randomly paired from within
a subplot (D
) and among subplots (D
) for hypothesis 1
and from within a plot (D
) and among plots (D
hypothesis 2 (Hardy and Senterre 2007, Baraloto et al.
2012; Data S2). Because the value of U
on the particular selection of a random subset of pairs of
individuals, we iterated this procedure 1,000 times and
generated a distribution of U
. By definition, U
take both positive and negative values, where U
indicates trait clustering and U
<0 trait overdispersion.
To assess the significance of U
, a community null
functional trait distribution (U
assembly as the null expectation was calculated using the
same procedure for each of the hypotheses, but breaking
down the community observed functional trait structure
by randomly shuffling taxonomic identities among indi-
viduals (T1 randomization sensu G€
otzenberger et al.
2016). As a result, the original community structure
remains unaltered because taxa richness and frequency
are fixed, but trait values are independently reshuffled
across taxa for each trait, thus not preserving the correla-
tion structure across traits. We selected this randomiza-
tion procedure over others because of its versatility:
although it is particularly suitable for detecting competi-
tive exclusion (via limiting similarity), it also performs
well in detecting environmental filtering (G€
FIG. 1. Location of the study sites. The two tropical montane Andean forests’elevational gradients in Ecuador and Peru (a, b).
Outlined areas represent Podocarpus National Park (c) and Rio Abiseo National Park (d), respectively. Ten plots of 0.1 ha were
established at each of three elevational belts (lower, intermediate, and upper) for each site, resulting in a total of 60 plots.
Article e03058; page 4 GUILLERMO BA
NARES-DE-DIOS ET AL. Ecology, Vol. 101, No. 7
et al. 2016). For each of the 1,000 U
tributions we extracted the mean value as a summary
statistic. Then, the distributions of the means of both the
observed and null U
were compared using a one-tailed t
test, with a critical significance level of a=0.05, because
tests for the null hypothesis were unidirectional (G€
berger et al. 2016). The whole analytic procedure was con-
ducted independently for each site and replicated for two
cohorts of co-occurring individuals: saplings
(DBH <10 cm) and adults (DBH ≥10 cm).
Assembly processes underlying community functional
For environmental characterization of each plot we
used the bioclimatic variables from CHELSA climate
data set (Karger et al. 2017). Mean annual temperature
(hereinafter MAT) was selected, as small changes in this
variable along our altitudinal gradients (spanning thermal
ranges between 9 and 12°C in Ecuador and Peru,
respectively; see Appendix S1: Table S1) are expected to
have a strong effect on species distribution, because in the
tropics species have evolved to have narrow thermal toler-
ances (Janzen 1967). In addition, MAT was highly corre-
lated (r>0.86) with most bioclimatic variables, both in
Ecuador and Peru. We used Mantel tests to analyze statis-
tically the correlation between the observed community
functional pattern for any pair of plots (U
hypothesis 2) and the plots’environmental distance, cal-
culated as the pairwise difference in MAT (Data S2). To
estimate the statistical significance of the correlation
between MAT and U
we used a Monte Carlo test,
permuting 200 times the elements of one of the distance
matrices while holding the other constant.
The climatic preferences of the most abundant taxa in
the lower and upper elevational belts (i.e., at both
extremes of the elevational gradient) were computed for
each one of the two study sites. To achieve this, we
defined the most abundant species as those with 10 or
more individuals within a single elevational belt
FIG. 2. Sampling design and the respective community functional pattern for testing the existence of different community
assembly processes at different spatial scales and the expected community functional patterns. (a) According to hypothesis 1, at
small spatial scale within plots, an overdispersed functional pattern occurs, which would be consistent with competitive exclusion.
Conversely, (b) in accordance with hypothesis 2, at larger spatial scale among plots, a clustered functional pattern arises, which
would suggest the existence of environmental filtering. U
is the mean functional trait Euclidean distance between pairs of co-oc-
curring individuals from distinct taxa randomly paired within subplots (D
) and among subplots (D
) for hypothesis 1 and within
) and among plots (D
) for hypothesis 2.
July 2020 COMMUNITY ASSEMBLY: PATTERNS AND PROCESSES Article e03058; page 5
(morphospecies excluded). In total we recognized 65 of
such species for the lower tropical montane forest belt
(henceforth, LTMF species) and 66 for the upper belt
(UTMF species) for Ecuador, whereas there were 26 and
27, respectively, for Peru (Appendix S1: Table S4; Data
S1). We then we retrieved occurrence data across the
Neotropics for each of those species from the Global
Biodiversity Information Facility (GBIF) and extracted
the bioclimatic information from CHELSA for the loca-
tions where species were reported. The climatic prefer-
ence of LTMF and UTMF species was defined as the
mean 1.96 standard deviation of the MAT from those
species’locations (Data S2). Finally, the significance of
the differences in climatic preferences between sets of
LTMF and UTMF species was estimated comparing the
mean MAT value for each set of species’locations with
a one-tailed ttest, using a significance level of a=0.05.
All analyses were conducted using the ‘vegan’R pack-
age (Oksanen et al. 2006).
As consequence of environmental filtering the species
adapted to certain climatic preferences would be filtered
out along the elevational gradient as climatic conditions
change. To explore this, we quantified the number of indi-
viduals of LTMF and UTMF species at each of the three
elevational belts (lower, intermediate, and upper) and
then analyzed how they changed with elevation in each of
the two study sites using generalized linear mixed models
(GLMMs) with a negative binomial error distribution (to
account for statistical overdispersion; not to be con-
founded with functional overdispersion). Mean elevation
at each belt, climatic preference (LTMF or UTMF) and
their interactions were used as fixed terms, whereas spe-
cies identity was used as a random factor. A random
slope structure was used for mean elevation, indicating
that the slope of the relationship between abundance and
elevation may change randomly among species. We built
all possible combinations of fixed and random factors.
Overall, we fitted 15 models, including null models for
both fixed and random effects (Appendix S1: Table S5),
using the R packages ‘glmmADMB”(Skaug et al. 2013)
and ‘MASS”(Venables and Ripley 2002). All models
were compared using Akaike’s information criterion, cor-
rected for small sample sizes (AIC
), with the R package
‘MuMIn’(Barton 2018; Data S2). Models with a differ-
ence in AIC
>2 indicate that the worse model had virtu-
ally no support and could be omitted.
Community assembly functional patterns
No statistically significant evidence (U
) of functional overdispersion was found at small spa-
tial scales for any of the three traits in either site (Fig. 3).
FIG. 3. Hypothesis 1: community functional distances distribution patterns at small spatial scale (within plot). Frequency of
distribution of U
(black) and U
(gray) values at two geographical sites (Ecuador, upper charts; Peru, lower charts) as
measured based on three functional traits (specific leaf area [SLA], left; leaf thickness [LT], center; wood density [WD], right) after
1,000 randomizations, in all six cases. There were no significant differences (a=0.05) between U
for any of the
traits at any of the sites.
Article e03058; page 6 GUILLERMO BA
NARES-DE-DIOS ET AL. Ecology, Vol. 101, No. 7
Thus, individuals within a subplot were not functionally
more different from the rest of individuals of the same
plot than expected by chance (U
same overall results were obtained when the analyses
were conducted independently for saplings and adults
(Appendix S1: Fig. S1).
Instead, when larger spatial scales were considered,
there was statistically significant evidence (U
,P≤0.01) of functional clustering for all
traits at both sites (Fig. 4). Thus, the individuals within
a plot were functionally more similar from the individu-
als of different plots than expected by chance (U
). The same overall results were obtained when
the analyses were independently conducted for saplings
and adults (Appendix S1: Fig. S2).
ASSEMBLY PROCESSES UNDERLYING COMMUNITY FUNCTIONAL
Trait clustering pattern (U
>0) among pairs of
plots increased when increasing differences in MAT
(Fig. 5). Those differences, as expected, were greater
when comparing plots from lower and upper elevational
belts (i.e., greater MAT differences) than when compar-
ing among plots within the same elevational belt (i.e.,
smaller or no MAT differences). The correlation
between MAT differences and trait clustering was posi-
tive and statistically significant (P≤0.01) for all three
functional traits at both sites. The only exception was
SLA in Peru, for which no significant variations in com-
munity trait pattern appeared in relation to differences
The climatic preferences of the LTMF and UTMF
species, defined using their MAT values across their
entire Neotropical distribution ranges, were clearly seg-
regated both in Ecuador and Peru (Fig. 6a, b). LTMF
species showed a mean temperature optimum of 22.8°C
in Ecuador and 24°C in Peru, whereas these optima were
15.5°and 18.2°C, respectively, for UTMF species. There
were statistically significant (P≤0.01) differences in
MAT for LTMF and UTMF species, both in Ecuador
The abundances of LTMF and UTMF species shifted
across the different elevational belts (Fig. 6c, d). Best-fit
models for both sites included the most complex struc-
ture for both fixed effects and the simplest for random
effects (Appendix S1: Table S5). Model predictions indi-
cated that LTMF species were significantly more abun-
dant at lower elevations than at the intermediate or,
especially, higher elevations, from which some species
FIG. 4. Hypothesis 2: community functional distances distribution patterns at large spatial scale (among plots). Frequency of
distribution of U
(black) and U
(gray) values at two geographical sites (Ecuador, upper charts; Peru, lower charts) as
measured based on three functional traits (specific leaf area [SLA], left; leaf thickness [LT], center; wood density [WD], right) after
1,000 randomizations, in all six cases. Differences between U
were significant (P≤0.01) for all traits at both sites.
July 2020 COMMUNITY ASSEMBLY: PATTERNS AND PROCESSES Article e03058; page 7
were absent. Conversely, UTMF species were signifi-
cantly more abundant at higher elevations than at the
intermediate or, especially, lower ones, where some were
Our study approaches a fundamental question in ecol-
ogy: identifying the ecological mechanisms shaping com-
munity assembly. Overall, we found that taking into
account spatial scale is key for detecting the functional
fingerprint of the underlying mechanisms driving com-
munity assembly. Although we found no evidence of
competitive exclusion at the smallest spatial scale, we
detected strong evidence of environmental filtering at
larger scales. In addition, further analyses conducted to
link the community observed functional pattern with its
underlying assembly process allows us to endorse the
role of environmental filtering for community assembly.
No evidence of competitive exclusion at small spatial
In this study we detected no evidence of competitive
exclusion at small spatial scale. Competitive exclusion
has been hypothesized to occur at small spatial scales,
where individuals from different species effectively com-
pete for local resources (Weiher and Keddy 1995b, Stoll
and Weiner 2000), thus a pattern of trait divergence is
expected as a consequence of limiting similarity (Wat-
kins and Wilson 2003, Stubbs and Wilson 2004). Never-
theless, as obvious as may seem, scale dependence has
sometimes been ignored. For instance, Baraloto et al.
(2012) rejected the importance of competitive exclusion
as an assembly process by comparing functional dis-
tances of species within 1-ha plots (D
) and among plots
separated by tens of kilometers (D
). In our opinion,
their results ought to be interpreted with caution for two
reasons. First, it makes little sense to test for competitive
exclusion between plant individuals that are very spa-
tially distant (e.g., up to 140 m apart from each other),
so are hardly competing for the same resources (e.g.,
light, soil nutrients). In relation to this, the scale at which
competition between species operates is certainly organ-
ism dependent, thus whereas sessile organisms like
plants mostly compete for key resources at small spatial
scales (up to a few tens of meters), motile organisms
such as birds or mammals can compete at much larger
scales (up to a few tens of kilometers). Consequently,
studies need to consider the spatial scales at which com-
munity assembly processes are most likely to operate in
relation to the group of organisms under study; for
example, if the organisms were plants, a checkerboard
pattern noticed at large spatial scales (Diamond 1975,
1982) could be misinterpreted as the effect of competi-
tive exclusion. Second, even if competitive exclusion
existed between individuals within such a large plot, its
trait overdispersion signal would be masked by the effect
of among-plots environmental differences on functional
To avoid spatial scale biases, we searched for evidence
of competitive exclusion at small spatial scale (i.e.,
within 10 910 m subplots). At this scale, co-occurring
individuals can be assumed to compete directly for the
same resources, and the effect of environmental filtering
on functional distance can be ruled out because environ-
mental conditions within a subplot are essentially the
same (de Bello et al. 2013b). However, conversely to our
expectation, we detected no evidence of a functional
overdispersion pattern resulting from competitive exclu-
sion for the traits we measured (Fig. 3). Instead, our
results suggest a random community assembly at this
small scale, although there could be cryptic nonrandom
dispersion with respect to traits that we did not measure
(Gallien 2017). In addition, because most of the assem-
bly processes operate more strongly at early stages of
FIG. 5. Plots pairwise functional and environmental comparison. The x-axes represent environmental differences (in terms of
mean annual temperature; MAT) between pairs of plots. The y-axes represent differences in observed functional traits patterns
) among plots for (a) specific leaf area [SLA], for (b) leaf thickness [LT], and (c) wood density [WD]. Trait clustering pattern
>0) increased positively as MAT differences increased. This correlation was significant (P≤0.01) for all traits at both
sites, excepting for SLA in Peru.
Article e03058; page 8 GUILLERMO BA
NARES-DE-DIOS ET AL. Ecology, Vol. 101, No. 7
plant life cycles (Green et al. 2014), a functional pattern
suggesting competitive exclusion may only become
revealed when only considering saplings in the commu-
nity, as they are most sensitive to competition (Falster
and Westoby 2003, Wagg et al. 2017). To test this
hypothesis, we checked whether functional overdisper-
sion emerged for saplings but disappeared for adults.
Again, no evidence of functional overdispersion was
found for either saplings or adults (Appendix S1:
Whereas some studies have also found no evidence of
functional overdispersion between co-occurring species
at small spatial scale (Schamp et al. 2008, Thompson
et al. 2010), others have. However, those studies that did
find functional overdispersion typically did not find it
for all the analyzed traits, and it was sometimes the case
that functional overdispersion and clustering were both
simultaneously reported within the same study system
(Cavender-Bares et al. 2004, Mason et al. 2007, Kraft
et al. 2008, Cornwell and Ackerly 2009, Swenson and
Enquist 2009, Kraft and Ackerly 2010, de Bello et al.
2013b). The few studies that systematically reported the
existence of limiting similarity (see Wilson 2007) were
based on evidence found in relatively low-diversity com-
munities, for example, sand dunes (Stubbs and Wilson
2004), lawns (Mason and Wilson 2006). or salt marshes
(Wilson and Stubbs 2012). Our results, in agreement
with the lack of consensus found in earlier studies, sug-
gest that in hyperdiverse plant communities, limiting
similarity may not be a paramount force driving com-
munity assembly (Grime 2006) because in hyperdiverse
systems, the functional hyperspace would be limited to
be parsed out into many discrete and differentiated func-
tional niches, each for one of the co-occurring species.
FIG. 6. Climatic preferences for the upper and lower tropical montane forest most abundant (N ≥10) species (upper tropical
montane forest [UTMF] and lower tropical montane forest [LTMF] species, respectively) and their abundances along the eleva-
tional gradients. Climatic preferences in terms of mean annual temperature (MAT) of the UTMF and LTMF sets of species were
segregated at the edges of the thermal gradient when the MAT values for those species’occurrences across their entire distribution
range in the Neotropics were considered, both in (a) Ecuador and (b) Peru. Dots represent the mean MAT and horizontal lines the
95 % confidence interval of the MAT for each set of species’climatic preferences. Difference in the means of MAT for LTMF and
UTMF species’climatic preferences was statistically significant (P≤0.01) in both sites. Shifts in UTMF and LTMF species’abun-
dances along the elevation gradients in (c) Ecuador and (d) Peru. Lines represent best fit model predictions with a 95% confidence
interval. Best-fit models for both sites included altitude, climatic preference, and their interaction as fixed terms, and species as a
July 2020 COMMUNITY ASSEMBLY: PATTERNS AND PROCESSES Article e03058; page 9
Evidence of environmental filtering clearly linked to
environmental heterogeneity at large spatial scale
In this study, we did find strong evidence of environ-
mental filtering at large spatial scale. Given the more abi-
otically homogeneous and restrictive environment
existing at small spatial scales, the potentially successful
functional strategies that allow the survival of commu-
nity members are narrowed, thus decreasing the role of
environmental filtering for community assembly. But at
large spatial scales that encompass different habitat con-
ditions, such as topography or edaphic variables, or that
expand environmental gradients, there is consensus on
the importance of environmental filtering. Under this
scenario, the selection of just the suitable set of traits that
allows species to thrive under certain environmental con-
ditions would result in a functional clustering pattern
(Fig. 4). Nevertheless, the reliability of clustering pat-
terns, in themselves widely accepted as indicators of envi-
ronmental filtering, has been questioned lately, because
some biotic processes can also render clustering patterns
(Sargent and Ackerly 2008). We agree with Kraft et al.
(2015) on their assertion that experimental manipula-
tions aimed to assess species’failure to establish and per-
sist in the absence of biotic interactions are the most
robust proof of environmental filtering sensu stricto. We
argue, however, that this is not only impractical in field
studies, especially at logistically challenging tropical
montane forests, it is also not necessarily meaningful
from an ecological perspective. Instead, we trust that,
according to Cadotte and Tucker (2017), as long as we
can correlate changes in community functional clustering
patterns, species abundances, or population growth with
the underlying shifts in environmental conditions, we can
infer and advocate for the existence of a sensu lato envi-
ronmental filtering process ongoing, regardless of simul-
taneously occurring biotic phenomena.
Our results show that only when large spatial scales
that truly encompass environmental differences are con-
sidered (e.g., among pairs of plots from different eleva-
tional belts and, thus, affected by notable MAT
differences) is environmental filtering revealed as an
overriding influence for community assembly. Thus, the
evidence of environmental filtering does not merely lay
in the traditionally admitted clustering pattern itself, but
in the fact that the pattern only arises when underlying
environmental heterogeneity exists (Fig. 5). In addition,
our study reveals how such environmental differences
cause changes in community features other than the
clustering pattern. For instance, for the most abundant
species in the lower and upper tropical montane forest—
LTMF and UTMF species, respectively—(1) their abun-
dances dramatically shift between elevations (Fig. 6c, d)
and (2) their climatic preferences are segregated at the
edges of the thermal gradient (Fig. 6a, b). Those facts
suggest that species distribution is mainly a consequence
of species abiotic preferences resulting from environmen-
tal filtering, and unlikely the result of other factors, such
as dispersal limitation, which is expected to play a negli-
gible role in the continuous, nonfragmented forests
within single river basins, as is the case in both our sites
(Young 1990, Pennington et al. 2010). In addition, the
fact that the climatic preferences of the LTMF and
UTMF species are maintained across their entire distri-
bution range in the Neotropics (Fig. 6a, b) reinforces
the role of environmental filtering as a broadly prevalent
mechanism for community assembly.
Thus far, indications of an environmental filtering fin-
gerprint on tropical community assembly across differ-
ent habitats have been reported mainly by considering
habitat differences as a surrogate for environmental dif-
ferences (e.g., topography in the Yasun
ı megaplot, Ecua-
dor, by Kraft et al. 2008, or forest age and geological
formation in Barro Colorado Island, Panama, by Gar-
zon-Lopez et al. 2014). Besides, identifying the effect of
environmental filtering has remained particularly chal-
lenging in species-rich forests, where, as a result of
stochastic dilution, the signal of deterministic assembly
processes may not be detectable, even if those processes
are operating (Wang et al. 2016). However, our study
provides robust scale and environmentally based evi-
dence supporting the importance of environmental fil-
tering for community assembly.
SUMMARY AND FURTHER PROSPECTS
The effect of environmental filtering in community
assembly has been traditionally inferred from trait clus-
tering patterns, as found by several studies targeted at
various systems and taxonomic groups (e.g., trees in
tropical forests, Baraloto et al. 2012; rockfishes in the
ocean, Ingram and Shurin 2009; bees in mountains, Pel-
lissier et al. 2013). However, in order to be properly
addressed, pertinent spatial scale–related implications
need to be taken into account (e.g., de Bello et al. 2009,
Swenson and Enquist 2009, M€
uller et al. 2013,
Garzon-Lopez et al. 2014, Mori et al. 2015). Our results
demonstrate a clear link between the pattern and the
mechanism by showing that the pattern is only revealed
when environmental differences exist, and by demon-
strating how those differences correlate with species’cli-
matic preferences, maintained across their entire
distribution range in the Neotropics, and abundances,
along elevation. This study thus contributes to empha-
sizing the importance of considering the implications of
spatial scale to detect the extent at which assembly
mechanisms act. In addition, it highlights the undeniable
role of environmental filtering in community assembly
and the usefulness of such a concept, demonstrating that
neither excluding biotic potentially confounding pro-
cesses nor identifying abiotic tolerance ranges are strictly
necessary for validating its effect. Future studies have
the challenge of advancing the discussion and shedding
light on the remaining details, such as whether the effect
of environmental filtering equally influences low- and
Article e03058; page 10 GUILLERMO BA
NARES-DE-DIOS ET AL. Ecology, Vol. 101, No. 7
GBD was funded through a PhD grant by the Spanish Min-
istry of Education (MINEDU; FPU14/05303) and an intern-
ship grant by Universidad Rey Juan Carlos (URJC Escuela
Internacional de Doctorado; Doctor Internacional 2017). The
study was supported through two grants from the Spanish Min-
istry of Economy and Competitiveness (MINECO; CGL2013-
45634-P, CGL2016-75414-P). We are indebted to those who
helped with fieldwork: Stalin Jap
on, Wilson Remacha, Percy
Malqui, “Rosho”Tamayo, Reinerio Ishuiza, Carlos Salas, Jos
anchez, Manuel Marca, Anselmo Vergaray, Gonzalo Ba~
Angel Delso, Julia Gonz
alez de Aledo, and Mara Paneghel. We
are especially thankful to Jorge Armijos, Alex Nina, Gabriel
Arellano, and Tolentino Cueva and the people in Los Alisos
(Pataz, Peru) for their invaluable help. In addition, we thank the
Ministry of Environment (MAE) in Ecuador and the National
Service of Natural Protected Areas (SERNANP) in Peru, in
ıctor Macedo, Vladimir Ram
ırez, Octavio Pecho,
and Jhonny Ramos. We extend our thanks to all the national
parks rangers that helped us, especially Rafael Gal
Heras, Percy Franco, and Guillermo Aguilar. Berardo Rojas
and Grober …Benites. Permits to work in protected areas were
granted by national authorities: Ecuador (MAE-DNB-CM-
2015-0016; No. 001-2019-IC-FLO-FAU-DPAZCH-UPN-VS/
MA) and Peru (001-2016-SERNANP-PNRA-JEF). Finally, we
wish to thank Jonathan Myers, Akira Mori, and an anonymous
referee for their very helpful comments on a manuscript draft.
Abrams, A. P. 1996. Limits to the similarity of competitors
under hierarchical lottery competition. American Naturalist
Ackerly, D. D., and W. K. Cornwell. 2007. A trait-based
approach to community assembly: Partitioning of species
trait values into within- and among-community components.
Ecology Letters 10:135–145.
Arellano, G., V. Cala, A. Fuentes, L. Cayola, P. M. Jorgensen,
and M. J. Mac
ıa. 2016. A standard protocol for woody plant
inventories and soil characterisation using temporary 0.1-ha
plots in tropical forests. Journal of Tropical Forest Science
Baraloto, C. et al. 2012. Using functional traits and phyloge-
netic trees to examine the assembly of tropical tree communi-
ties. Journal of Ecology 100:690–701.
Baraloto, C., C. E. T. Paine, L. Poorter, J. Beauchene, D. Bonal,
A. M. Domenach, B. H
erault, S. Pati~
no, J. C. Roggy, and J.
Chave. 2010. Decoupled leaf and stem economics in rain for-
est trees. Ecology Letters 13:1338–1347.
Barton, K.2018. MuMIn: Multi-model inference. R package.
Bazzaz, F. A. 1991. Habitat selection in plants. The American
Belyea, L. R., and J. Lancaster. 1999. Assembly rules within a
contingent ecology. Oikos 86:402–416.
Burns, J. H., and S. Y. Strauss. 2011. More closely related spe-
cies are more ecologically similar in an experimental test. Pro-
ceedings of the National Academy of Sciences of the United
States of America 108:5302–5307.
Cadotte, M. W., and C. M. Tucker. 2017. Should environmental
filtering be abandoned? Trends in Ecology and Evolution
Cavender-Bares, J., D. D. Ackerly, D. A. Baum, and F. A. Baz-
zaz. 2004. Phylogenetic overdispersion in Floridian Oak com-
munities. American Naturalist 163:823–843.
Cavender-Bares, J., K. H. Kozak, P. V. A. Fine, and S. W. Kem-
bel. 2009. The merging of community ecology and phyloge-
netic biology. Ecology Letters 12:693–715.
Cayuela, L., I. Granzow-de la Cerda, F. S. Albuquerque, and D.
J. Golicher. 2012. TAXONSTAND: An R package for species
names standardisation in vegetation databases. Methods in
Ecology and Evolution 3:1078–1083.
Cayuela, L., A. Stein, and J. Oksanen.2017. Taxonstand: taxo-
nomic standardization of plant species names. R package.
Chalmandrier, L. et al. 2017. Spatial scale and intraspecific trait
variability mediate assembly rules in alpine grasslands. Jour-
nal of Ecology 105:277–287.
Chase, J. M. 2014. Spatial scale resolves the niche versus neutral
theory debate. Journal of Vegetation Science 25:319–322.
Chesson, P. 2000. The maintenance of species diversity. Annual
Review of Ecology and Systematics 31:343–366.
Clements, F. E. 1916. Plant succession: an analysis of the devel-
opment of vegetation. Carnegie Institution of Washington,
Washington, D.C., USA.
Colwell, R. K., and D. Winkler.1984. A null model for null
models in biogeography. Pages 344–359 in D. R. Strong, D.
Simberloff, L. G. Abele, and A. B. Thistle, editors. Ecological
communities: conceptual issues and the evidence. Princeton
University Press, Princeton, New Jersey, USA.
Connor, E. F., and D. Simberloff. 1979. The assembly of species
communities: chance or competition? Ecology 60:1132–1140.
Cornelissen, J. H. C. et al. 2003. A handbook of protocols for
standardised and easy measurement of plant functional traits
worldwide. Australian Journal of Botany 51:335–380.
Cornwell, W. K., and D. D. Ackerly. 2009. Community assem-
bly and shifts in the distribution of functional trait values
across an environmental gradient in coastal California. Eco-
logical Monographs 79:109–126.
Cornwell, W. K., D. Schwilk, and D. D. Ackerly. 2006. A trait-
based test for habitat filtering: convex hull volume. Ecology
Dayan, T., and D. Simberloff. 2005. Ecological and community-
wide character displacement: the next generation. Ecology
de Bello, F., S. Lavorel, S. Lavergne, C. H. Albert, I. Boulan-
geat, F. Mazel, and W. Thuiller. 2013a. Hierarchical effects of
environmental filters on the functional structure of plant
communities: a case study in the French Alps. Ecography
de Bello, F., W. Thuiller, J. Lep
Macek, M. T. Sebasti
a, and S. Lavorel. 2009. Partitioning of
functional diversity reveals the scale and extent of trait conver-
gence and divergence. Journal of Vegetation Science 20:475–486.
de Bello, F., M. Vandewalle, T. Reitalu, J. Lep
s, H. C. Prentice,
S. Lavorel, and M. T. Sykes. 2013b. Evidence for scale- and
disturbance-dependent trait assembly patterns in dry semi-
natural grasslands. Journal of Ecology 101:1237–1244.
Diamond, J. M. 1975. Assembly of species communities. Pages
342–444 in L. Cody and J. Diamond, editors. Ecology and
evolution of communities. Harvard University Press, Cam-
bridge, Massachusetts, USA.
Diamond, J. M. 1982. Effect of species pool size on species
occurrence frequencies: musical chairs on islands. Proceed-
ings of the National Academy of Sciences of the United
States of America 79:2420–2424.
ıaz, S., M. Cabido, and F. Casanoves. 1998. Plant functional
traits and environmental filters at a regional scale. Journal of
Vegetation Science 9:113–122.
Falster, D. S., and M. Westoby. 2003. Plant height and evolu-
tionary games. Trends in Ecology and Evolution 18:337–343.
July 2020 COMMUNITY ASSEMBLY: PATTERNS AND PROCESSES Article e03058; page 11
Fukami, T., T. M. Bezemer, S. R. Mortimer, and W. H. Van Der
Putten. 2005. Species divergence and trait convergence in
experimental plant community assembly. Ecology Letters
Funk, J. L., J. E. Larson, G. M. Ames, B. J. Butterfield, J.
Cavender-Bares, J. Firn, D. C. Laughlin, A. E. Sutton-Grier,
L. Williams, and J. Wright. 2017. Revisiting the holy grail:
using plant functional traits to understand ecological pro-
cesses. Biological Reviews 92:1156–1173.
Gallien, L. 2017. Intransitive competition and its effects on
community functional diversity. Oikos 126:615–623.
Garzon-Lopez, C. X., P. A. Jansen, S. A. Bohlman, A. Ordo~
and H. Olff. 2014. Effects of sampling scale on patterns of
habitat association in tropical trees. Journal of Vegetation
Gause, G. F. 1934. The struggle for existence. Williams and Wil-
liams, Baltimore, Maryland, USA.
Godoy, O., N. J. B. Kraft, and J. M. Levine. 2014. Phylogenetic
relatedness and the determinants of competitive outcomes.
Ecology Letters 17:836–844.
Gotelli, N. J., and G. R. Graves.1996. Null models in ecology.
Smithsonian Institution Press, Washington, D.C., USA.
otzenberger, L. et al. 2012. Ecological assembly rules in plant
communities—approaches, patterns and prospects. Biological
otzenberger, L., Z. Botta-Duk
at, J. Lep
s, M. P€
artel, M. Zobel,
and F. de Bello. 2016. Which randomizations detect conver-
gence and divergence in trait-based community assembly? A
test of commonly used null models. Journal of Vegetation
Green, P. T., K. E. Harms, and J. H. Connell. 2014. Nonran-
dom, diversifying processes are disproportionately strong in
the smallest size classes of a tropical forest. Proceedings of
the National Academy of Sciences of the United States of
Grime, J. P. 2006. Trait convergence and trait divergence in
herbaceous plant communities: Mechanisms and conse-
quences. Journal of Vegetation Science 17:255–260.
Hardy, O. J., and B. Senterre. 2007. Characterizing the phyloge-
netic structure of communities by an additive partitioning of
phylogenetic diversity. Journal of Ecology 95:493–506.
HilleRisLambers, J., P. B. Adler, W. S. Harpole, J. M. Levine,
and M. M. Mayfield. 2012. Rethinking community assembly
through the lens of coexistence theory. Annual Review of
Ecology, Evolution, and Systematics 43:227–248.
Hubbell, S. P. 2001. The unified neutral theory of biodiversity
and biogeography. Princeton University Press, Princeton,
New Jersey, USA.
Ingram, T., and J. B. Shurin. 2009. Trait-based assembly and
phylogenetic structure in northeast Pacific rockfish assem-
blages. Ecology 90:2444–2453.
Janzen, D. H. 1967. Why mountain passes are higher in the
tropics. American Naturalist 101:233–249.
Karger, D. N., O. Conrad, J. B€
ohner, T. Kawohl, H. Kreft, R.
W. Soria-auza, N. E. Zimmermann, H. P. Linder, and M.
Kessler. 2017. Climatologies at high resolution for the earth’
s land surface areas. Scientific Data 4:170122.
Keddy, P. A. 1992. Assembly and response rules: two goals for
predictive community ecology. Journal of Vegetation Science
Kneitel, J. M., and J. M. Chase. 2004. Trade-offs in community
ecology: linking spatial scales and species coexistence. Ecol-
ogy Letters 7:69–80.
Kraft, N. J. B., and D. D. Ackerly. 2010. Functional trait and
phylogenetic tests of community assembly across spatial scales
in an Amazonian forest. Ecological Monographs 80:401–422.
Kraft, N. J. B., and D. D. Ackerly. 2014. Assembly of plant com-
munities. InR. K. Monson, editor. Ecology and the environ-
ment, the plant sciences 8. Springer-Verlag, Berlin, Germany.
Kraft, N. J. B., P. B. Adler, O. Godoy, E. C. James, S. Fuller,
and J. M. Levine. 2015. Community assembly, coexistence
and the environmental filtering metaphor. Functional Ecol-
Kraft, N. J. B., R. Valencia, and D. D. Ackerly. 2008. Func-
tional traits and niche-based tree community assembly in an
Amazonian forest. Science 322:580–582.
Le Bagousse-Pinguet, Y., N. Gross, F. T. Maestre, V. Maire, F.
de Bello, C. R. Fonseca, J. Kattge, E. Valencia, J. Leps, and P.
Liancourt. 2017. Testing the environmental filtering concept
in global drylands. Journal of Ecology 105:1058–1069.
Lebrija-Trejos, E., E. A. P
ıa, J. A. Meave, F. Bongers,
and L. Poorter. 2010. Functional traits and environmental fil-
tering drive community assembly in a species-rich tropical
system. Ecology 91:386–398.
Levin, S. A. 1992. The problem of pattern and scale in ecology.
opez-Angulo, J., N. G. Swenson, L. A. Cavieres, and A.
Escudero. 2018. Interactions between abiotic gradients deter-
mine functional and phylogenetic diversity patterns in
Mediterranean-type climate mountains in the Andes. Journal
of Vegetation Science 29:245–254.
MacArthur, R., and R. Levins. 1967. The limiting similarity,
convergence and divergence of coexisting species. American
Mason, N. W. H., C. Lanoisel
ee, D. Mouillot, P. Irz, and C.
Argillier. 2007. Functional characters combined with null
models reveal inconsistency in mechanisms of species turn-
over in lacustrine fish communities. Oecologia 153:441–452.
Mason, N. W. H., and J. B. Wilson. 2006. Mechanisms of spe-
cies coexistence in a lawn community: Mutual corroboration
between two independent assembly rules. Community Ecol-
Mayfield, M. M., and J. M. Levine. 2010. Opposing effects of
competitive exclusion on the phylogenetic structure of com-
munities. Ecology Letters 13:1085–1093.
McGill, B. J. 2010. Matters of scale. Science 328:575–576.
McGill, B. J., B. J. Enquist, E. Weiher, and M. Westoby. 2006.
Rebuilding community ecology from functional traits. Trends
in Ecology and Evolution 21:178–185.
Mori, A. S., A. T. Ota, S. Fujii, T. Seino, D. Kabeya, T. Oka-
moto, M. T. Ito, N. Kaneko, and M. Hasegawa. 2015. Biotic
homogenization and differentiation of soil faunal communi-
ties in the production forest landscape: taxonomic and func-
tional perspectives. Oecologia 177:533–544.
uller, T. et al 2013. Scale decisions can reverse conclu-
sions on community assembly processes. Global Ecology and
Narwani, A., M. A. Alexandrou, T. H. Oakley, I. T. Carrol, and
B. J. Cardinale. 2013. Experimental evidence that evolution-
ary relatedness does not affect the ecological mechanisms of
coexistence in freshwater green algae. Ecology Letters
Oksanen, J., R. Kindt, P. Legendre, B. O’Hara, and H. H. Ste-
vens. 2006. Vegan: community ecology package. R package.
Parker, I. M., M. Saunders, M. Bontrager, A. P. Weitz, R. Hen-
dricks, R. Magarey, K. Suiter, and G. S. Gilbert. 2015. Phylo-
genetic structure and host abundance drive disease pressure
in communities. Nature 520:542–544.
Pellissier, L., J. N. Pradervand, P. H. Williams, G. Litsios, D.
Cherix, and A. Guisan. 2013. Phylogenetic relatedness and
proboscis length contribute to structuring bumblebee
Article e03058; page 12 GUILLERMO BA
NARES-DE-DIOS ET AL. Ecology, Vol. 101, No. 7
communities in the extremes of abiotic and biotic gradients.
Global Ecology and Biogeography 22:577–585.
Pennington, R. T., M. Lavin, T. Sarkinen, G. P. Lewis, B. B.
Klitgaard, and C. E. Hughes. 2010. Contrasting plant diversi-
fication histories within the Andean biodiversity hotspot.
Proceedings of the National Academy of Sciences of the Uni-
ted States of America 107:13783–13787.
Sargent, R. D., and D. D. Ackerly. 2008. Plant–pollinator inter-
actions and the assembly of plant communities. Trends in
Ecology and Evolution 23:123–130.
Schamp, B. S., J. Chau, and L. W. Aarssen. 2008. Dispersion of
traits related to competitive ability in an old-field plant com-
munity. Journal of Ecology 96:204–212.
Silvertown, J. 2004. Plant coexistence and the niche. Trends in
Ecology and Evolution 19:605–611.
Skaug, H., D. Fournier,A. Nielsen, A. Magnusson, and B. Bolker.
2013. Generalized linear mixed models using “AD Model
Snyder, R. E., and P. Chesson. 2004. How the spatial scales of
dispersal, competition, and environmental heterogeneity inter-
act to affect coexistence. American Naturalist 164:633–650.
Stoll, P., and J. Weiner.2000. A neighborhood view of interac-
tions among individual plants. Pages 11–27 in U. Dieckmann,
R. Law, and J. Metz, editors. The geometry of ecological
interactions: simplifying spatial complexity. Cambridge
University Press, Cambridge, UK.
Stubbs, W. J., and J. B. Wilson. 2004. Evidence for limiting simi-
larity in a sand dune community. Journal of Ecology 92:557–
Sutherland, W. J. et al. 2013. Identification of 100 fundamental
ecological questions. Journal of Ecology 101:58–67.
Swenson, N. G. 2012. The functional ecology and diversity of
tropical tree assemblages through space and time: from local
to regional and from traits to transcriptomes. ISRN Forestry
Swenson, N. G., and B. J. Enquist. 2008. The relationship
between stem and branch wood specific gravity and the abil-
ity of each measure to predict leaf area. American Journal of
Swenson, N. G., and B. J. Enquist. 2009. Opposing assembly
mechanisms in a Neotropical dry forest: Implications for phy-
logenetic and functional community ecology. Ecology
Swenson, N. G., B. J. Enquist, J. Thompson, and J. K. Zimmer-
man. 2007. The influence of spatial and size scale on phyloge-
netic relatedness in tropical forest communities. Ecology
Thiers, W. n.d. Index Herbariorum: A global diectory of public
herbaria and associated staff. New York Botanical Garden’s
Virtual Herbarium. http://sweetgum.nybg.org/science/ih/
Thompson, K., O. L. Petchey, A. P. Askew, N. P. Dunnett, A. P.
Beckerman, and A. J. Willis. 2010. Little evidence for limiting
similarity in a long-term study of a roadside plant commu-
nity. Journal of Ecology 98:480–487.
Uriarte, M. 2000. Interactions between goldenrod (Solidago
altissima L.) and its insect herbivore (Trirhabda virgata) over
the course of succession. Oecologia 122:521–528.
Venables, W., and B. Ripley. 2002. Modern applied statistics
with S. Springer, New York, New York, USA.
Violle, C., M.-L. Navas, D. Vile, E. Kazakou, C. Fortunel, I.
Hummel, and E. Garnier. 2007. Let the concept of trait be
Wagg, C. et al. 2017. Functional trait dissimilarity drives both
species complementarity and competitive disparity. Func-
tional Ecology 31:2320–2329.
Wang, X. et al. 2016. Stochastic dilution effects weaken deter-
ministic effects of niche-based processes in species rich for-
ests. Ecology 97:347–360.
Watkins, A. J., and J. B. Wilson. 2003. Local texture conver-
gence: A new approach to seeking assembly rules. Oikos
Webb, C. O., D. D. Ackerly, M. A. McPeek, and M. J. Dono-
ghue. 2002. Phylogenies and community ecology. Annual
Review of Ecology and Systematics 33:475–505.
Weiher, E., and P. A. Keddy. 1995a. The assembly of experimen-
tal wetland plant communities. Oikos 73:323–335.
Weiher, E., and P. A. Keddy. 1995b. Assembly rules, null models
and trait dispersion: new questions from old patterns. Oikos
Westoby, M., and I. J. Wright. 2006. Land-plant ecology on the
basis of functional traits. Trends in Ecology and Evolution
Whittaker, R. J., K. J. Willis, and R. Field. 2001. Scale and spe-
cies richness: towards a general, hierarchical theory of species
diversity. Journal of Biogeography 28:453–470.
Wiens, J. A. 1989. Spatial scaling in ecology. Functional Ecol-
Wilson, J. B. 2007. Trait-divergence assembly rules have been
demonstrated: limiting similarity lives!A reply to Grime.
Journal of Vegetation Science 18:451–452.
Wilson, J. B., and W. J. Stubbs. 2012. Evidence for assembly
rules: limiting similarity within a saltmarsh. Journal of Ecol-
Wright, I. J. et al. 2004. The worldwide leaf economics spec-
trum. Nature 428:821–827.
Wright, S. J. 2002. Plant diversity in tropical forests: A review of
mechanisms of species coexistence. Oecologia 130:1–14.
Young, K. R. 1990. Dispersal of Styrax ovatus seeds by the
spectacled bear (Tremarctos ornatus). Vida Silvestre Neotrop-
Additional supporting information may be found in the online version of this article at http://onlinelibrary.wiley.com/doi/
July 2020 COMMUNITY ASSEMBLY: PATTERNS AND PROCESSES Article e03058; page 13