ArticlePDF Available

Functional determinants of forest recruitment over broad scales

Authors:

Abstract and Figures

AimThe drivers of tree recruitment over large spatial scales remain unexplored. Here, we ask whether species potential for recruitment and the strength of density-dependent processes, both inferred from species relative abundances, show emerging patterns that can be explained upon the basis of information about climate and functional traits. LocationEastern forests of the USA. Methods We document the geographical distributions and magnitudes of seedling recruitment and the strength of density dependence and conspecific density dependence for the forests of the eastern USA spanning >1.2 million km(2) across 88,854 local communities comprising 164 tree species. We also compiled climatic variables and 16 traits representing several important ecological axes of tree functional strategies to assess which factors were most strongly associated with the emerging broad-scale spatial patterns. ResultsStrong geographical variation in the potential for seedling recruitment and a latitudinal change from negative to positive density dependence moving northward were associated with adaptation to seasonal freezing temperatures and seed size. Wood density and leaf nitrogen, in contrast, were related to the magnitude of the negative density dependence and conspecific density dependence, respectively, which were prevalent over most of the region. Main conclusionsOur results provide strong evidence that tree recruitment and the strength of density-dependent processes have broad-scale patterns that can be explained by a few key species functional traits.
Content may be subject to copyright.
RESEARCH
PAPER
Functional determinants of forest
recruitment over broad scales
Oscar Godoy1*, Marta Rueda2and Bradford A. Hawkins2
1Department of Ecology, Evolution and Marine
Biology, University of California, Santa
Barbara, CA 93106, USA, 2Department of
Ecology and Evolutionary Biology, University
of California, Irvine, CA 92612, USA
ABSTRACT
Aim The drivers of tree recruitment over large spatial scales remain unexplored.
Here, we ask whether species potential for recruitment and the strength of density-
dependent processes, both inferred from species relative abundances, show emerg-
ing patterns that can be explained upon the basis of information about climate and
functional traits.
Location Eastern forests of the USA.
Methods We document the geographical distributions and magnitudes of seed-
ling recruitment and the strength of density dependence and conspecific density
dependence for the forests of the eastern USA spanning >1.2 million km2across
88,854 local communities comprising 164 tree species. We also compiled climatic
variables and 16 traits representing several important ecological axes of tree func-
tional strategies to assess which factors were most strongly associated with the
emerging broad-scale spatial patterns.
Results Strong geographical variation in the potential for seedling recruitment
and a latitudinal change from negative to positive density dependence moving
northward were associated with adaptation to seasonal freezing temperatures and
seed size. Wood density and leaf nitrogen, in contrast, were related to the magnitude
of the negative density dependence and conspecific density dependence, respec-
tively, which were prevalent over most of the region.
Main conclusions Our results provide strong evidence that tree recruitment and
the strength of density-dependent processes have broad-scale patterns that can be
explained by a few key species functional traits.
Keywords
Conspecific density dependence, density independence, frost phenology,
functional traits, leaf N, North America, positive interactions, species
abundances, wood density.
*Correspondence: Oscar Godoy.
E-mail: ogodoy@irnas.csic.es
Present address: Instituto de Recursos
Naturales y Agrobiología de Sevilla (IRNAS),
CSIC, PO Box 1052, Sevilla E-41080, Spain.
INTRODUCTION
The patterns must be understood as emerging from the col-
lective behaviors of large ensembles of smaller scale units
(Levin, 1992).
Ecologists and foresters have studied tree recruitment for
decades, given its importance for understanding forest dynamics
(Clark et al., 1998, 2010; Rees et al., 2001; Comita et al., 2010),
the maintenance of forest diversity (Bolker & Pacala, 1999;
Harms et al., 2000; Johnson et al., 2012) and, more recently, the
responses of species to climatic change (Peñuelas & Boada, 2003;
Zhu et al., 2012). Much effort has focused on local scales, but
there are few studies of biogeographical patterns emerging from
local assemblages (Johnson et al., 2012; Coll et al., 2013;
Carnicer et al., 2014; Ruiz-Benito et al., 2014). This severely
limits our understanding of the climatic and functional factors
driving patterns of tree species recruitment at scales relevant to
global change.
Variation in the potential for recruitment among tree species
can arise from endogenous factors (i.e. density-independent
processes), such as genetically based differences in fecundity,
masting, the tolerance of seedlings to drought and cold or age at
maturity (Rejmánek & Richardson, 1996; Clark et al., 1998;
Herrera et al., 1998; Mueller et al., 2005). This species potential
bs_bs_banner
Global Ecology and Biogeography, (Global Ecol. Biogeogr.) (2015) 24, 192–202
DOI: 10.1111/geb.12241
192 © 2014 John Wiley & Sons Ltd http://wileyonlinelibrary.com/journal/geb
for recruitment can also be strongly affected by density-
dependent processes, either from interactions with neighbours
for limiting resources, such as light and soil nutrients, or more
specifically from conspecific neighbours due to Janzen–Connell
effects of pathogens and seed and seedling predators (Janzen,
1970; Connell, 1971; Comita et al., 2010; Rees, 2013). Impor-
tantly, species with high relative recruitment often dominate
local communities. This can occur due to the production of
copious offspring, which is not sensitive to neighbours (Clark
et al., 1998), and from a positive relationship between offspring
survival and the number of conspecific neighbours (McIntire &
Fajardo, 2011). In contrast, a more diverse forest community is
maintained when, inter alia, species limit the establishment of
their own seedlings more than do the species with which they
interact (Chesson, 2000; Comita et al., 2010; Johnson et al.,
2012).
Although both species potential for recruitment and the
strength of density-dependent processes strongly interact with
abiotic and biotic drivers (e.g. the availability of light and
moisture or pathogen/herbivore populations) (Borchert et al.,
1989; Ribbens et al., 1994), which may result in high variability
of recruitment at local scales (Clark et al., 2010), we hypoth-
esize that they should also have broad-scale patterns related to
climatic variation. These emerging geographical patterns may
occur from the single effect of limiting conditions such as
drought or freezing as well as from the correlated effect that
climate has on shaping the geographical structure of important
species traits influencing fecundity, germination and seedling
mortality. For instance, we hypothesize that the trade-off
between seed size and seed number (Smith & Fretwell, 1974),
recently demonstrated for trees (Adler et al., 2014), will
translate into a geographical pattern of higher fecundity and
thus higher recruitment at northern latitudes and higher
elevations where communities are composed of tree species
producing smaller seeds (Moles et al., 2007; Hawkins et al.,
2014).
We also hypothesize that an increase in the density of neigh-
bours will reduce recruitment (Lambers et al., 2002; Comita
et al., 2010), but interactive effects may occur with climatic con-
ditions, shifting from negative to positive with increasing envi-
ronmental stress (Callaway et al., 2002; McIntire & Fajardo,
2011). For example,earlier work suggests that wood density may
determine the geographical pattern of the strength of density
dependence (Enquist et al., 1999; Swenson & Enquist, 2007;
Chave et al., 2009; Adler et al., 2014). Species with dense wood,
which in North American forests are distributed in southern
latitudes and at lower elevations, tend to be larger, have longer
life spans and survive better in the face of drought or insect/
pathogen attack, which may produce stronger negative effects of
neighbours for longer. Other functional traits related to resource
acquisition, such as height (for light) or rooting depth (for
water), can also determine the strength of density dependence
when resources are limiting (Sterck et al., 2011).
Recent work has shown that the strength of conspecific
density dependence in North American forests decreases with
latitude (Johnson et al., 2012), perhaps because drivers limiting
recruitment, such as host-specific herbivores and pathogens,are
themselves limited by climate (Janzen, 1970; Connell, 1971;
Harms et al., 2000), although the only empirical test of this
hypothesis found no support (Lambers et al., 2002). This sug-
gests that the geographical pattern of conspecific density
dependence should be similar to the overall density dependence.
It is less clear what the functional determinants of the strength
of conspecific density dependence are, as it tends to vary with
shade tolerance (Kobe & Vriesendorp, 2011). Shade tolerance is
a complex attribute that can be mechanistically driven by an
array of functional traits, including height, wood density, seed
size, leaf area and leaf N (Kitajima, 1994; Niinemets, 1997;
Hewitt, 1998). Such complex interactions among suites of traits
make it difficult to predict the distribution of phenotypes across
space.
In this paper we estimate the species potential for recruit-
ment arising from density-independent processes (hereafter
‘recruitment potential’, RP), and the strength of density
dependence among all neighbours (hereafter ‘density depend-
ence’, DD) and among conspecific neighbours only (hereafter
‘conspecific density dependence’, CDD) to document the com-
munity geographical patterns for the forests of the eastern
United States. We considered in total 164 tree species occurring
across 88,854 sites (c. 1.27 million km2). These demographic
variables were then analysed using a machine learning method
for regression (Random Forest) with respect to four climatic
variables accounting for variation in temperature and precipi-
tation, and with 16 functional traits (Table 1) representing
several important ecological axes of the strategies of woody
plants in terms of leaf economics, growth allocation, resource
acquisition and regeneration (Westoby et al., 2002; Westoby &
Wright, 2006; Kraft et al., 2008). We focus on three questions:
(1) Do local species assemblages show geographic patterns of
recruitment potential and density-dependent processes? (2) If
so, can the geographic patterns of recruitment potential and
density dependence be explained by climate and species traits?
(3) What is the relative importance of climatic versus func-
tional factors?
METHODS
Extracting information from the Forest Inventory
and Analysis database
Our study centred on the forests of the USA delimited by the
Northern and Southern Region units of the US Forest Service’s
Forest Inventory and Analysis (FIA). These forests extend
from Florida to Maine and from the Atlantic Coast to the
Kansas–Colorado line. We downloaded from the FIA (http://
www.fia.fs.fed.us; accessed February 2012) data collected
between 2005 and 2010, corresponding to the most recently
updated 5-year cyclical inventory. The FIA protocol is designed
to record, per site and per species, the number of trees larger
than 12.7 cm diameter at breast height (d.b.h.) in four subplots
(168.3 m2) 36.6 m apart, and the number of seedlings in
contiguous nested microplots (13.5 m2). The FIA defines as
Forest recruitment and species traits
Global Ecology and Biogeography,24, 192–202, © 2014 John Wiley & Sons Ltd 193
seedlings those individuals smaller than 2.54 cm d.b.h. and taller
than 30.5 cm for angiosperms and 12.2 cm for gymnosperms.
We used Access queries per state to select only those sites corre-
sponding to natural stands (i.e. excluding tree plantations and
orchards), without human disturbance and without the occur-
rence of fire in the 5 years prior to being surveyed. We then used
the FIPS code (the species code used by the FIA) to remove those
entries corresponding to broad tree classifications (e.g. ever-
green, deciduous) and genera, and to merge species under cur-
rently recognized names. We used The Plant List (http://
www.theplantlist.org; accessed March 2012) as the source of
information on synonyms. These selection criteria rendered a
total of 88,854 sites containing 164 species.
Estimation of recruitment potential and
density-dependent effects
We followed an analytical approach similar to that of Johnson
et al. (2012). First, we used each of the four subplot–microplot
pairs per site as our unit of analysis (replicates). Second, accord-
ing to the scatterplots of the number of seedlings per focal
species (Si) as a function of the total number of adult neighbours
(T), we fitted an exponential function using maximum likeli-
hood methods (R version 2.13.2, function ‘optim,’ method
‘L-BFGS-B’) with the specific form:
Sa
ii
bT
i
=e. (1)
Here, aiis the potential number of seedlings recruited per
species in the absence of interaction with adult neighbours (RP)
(i.e. recruitment from density-independent effects) and biis the
inflection curve parameter, which indicates the direction and
the magnitude of the effect of the density of adult neighbours on
RP. Negative values of biindicate that the number of seedlings
decreases exponentially as the number of adults increases (i.e.
negative, or direct, density dependence) whereas positive values
indicate the reverse (i.e. positive, or inverse,density dependence)
(Appendix S1 in Supporting Information). For the fitting
process, however, we selected a negative binomial error struc-
ture, giving the greatest heterogeneity observed in the increase of
the variance of the data with respect to the mean across species
(overdispersion parameter kranging from 0.318 to 7835,
median =1.399; smaller values of kindicate greater heterogene-
ity) (Bolker, 2008). We followed this procedure to assess the
average species response to the overall density-dependent effect
of conspecific and heterospecific neighbours (DD) as well as to
the separate effect of conspecific neighbours (CDD). For CDD,
we selected sites containing only seedlings and adults of a single
focal species without the presence of other species. The main
advantage of this function is to obtain a joint fit for ai(bounded
to be >1) and bi(set to be free). This also makes values of the
Table 1 Environmental and functional traits predictors included in the analyses.
Type Name nDefinition Meaning
Climatic Annual precipitation (mm)
Summer precipitation (mm)
Max. temperature (°C)
Min. temperature (°C)
Species traits Anaerobic tolerance 141 Relative tolerance to anaerobic soil conditions Stress tolerance
CaCO3tolerance 141 Relative tolerance to calcareous soils Stress tolerance
Dispersal mode 159 Unassisted, animal or wind Disturbance and dispersal
Drought tolerance 141 Relative species tolerance to drought conditions
compared with other species with the same growth
habit from the same geographical region
Stress tolerance
Fire tolerance 139 Relative ability to resprout, regrow or re-establish from
residual seed after a fire
Regeneration, disturbance and
resource use
Growth rate 137 Growth rate after successful establishment relative to
other species with the same growth habit
Resource use
Max. height (m) 164 Light acquisition
Leaf nitrogen (Nmass %) 106 Leaf economics – resource capture
Leaf phosphorus (Pmass %) 82 Leaf economics – resource
Max. trunk diameter (m) 144 Light acquisition
Min. frost-free days 140 The minimum average number of frost-free days
within the species’ geographical range
Stress tolerance
Min root depth (cm) 141 The minimum depth of soil required for good growth Stress tolerance
Leaf phenology 164 Evergreen versus deciduous Resource use
Seed mass (mg) 155 Regeneration
Shade tolerance 143 Relative tolerance to shade conditions Light acquisition
Wood density (mg cm–3) 113 Resource use, resistance to pathogens
nrepresents the number of species with information for each trait. The total number of species in the database is 164. Source references are included
in Appendix S3.
O. Godoy et al.
Global Ecology and Biogeography,24, 192–202, © 2014 John Wiley & Sons Ltd194
inflection curve (bi) sensitive to those fitted to the intercept (ai).
Therefore, we confirmed that differences in the strength of
average DD versus average CDD were not due to differences in
average RP (the correlation between y-intercepts of DD and
CDD fits was r=0.809, P<0.001). For both DD and CDD we
set the threshold of the number of replicates per species to 10
subplot–microplot pairs to perform the fitting, and finally we
did not divide the data into regional subunits because: (1) we
needed to match the resolution of trait information, which is
available as a single value per species (see Climatic and species
trait information), and (2) we needed a single value per species
for phylogenetic comparative analyses to test whether the
obtained parameters can be considered a species characteristic
for mapping (see next subsection).
Phylogenetic analyses testing whether recruitment
and density-dependent effects are species specific
We asked whether RP, DD and CDD values can be considered a
species characteristic or are evolutionarily labile. This was a
crucial step in identifying the type of information necessary to
explore emerging geographical patterns of recruitment potential
and density-dependent effects. The rationale is that if they are
labile or weakly conserved they are likely be a product of the
local environment; that is, two closely related species could have
either similar or very different values. Thus, environmental data
would be needed to map geographical patterns of RP, DD and
CDD. In contrast, if they are conserved we can assume that they
are a species characteristic and closely related species will tend to
have similar values while values for disparate lineages will tend
on average to diverge. In such a case, we would consider species
distributions rather than environment a better approach for
exploring geographical patterns.
We performed two types of phylogenetic comparative
analyses: phylogenetic signal representation curves (PSR)
(Diniz-Filho et al., 2012) and the fitting by phylogenetic gener-
alized least squares (pGLS) of three models of evolution given
our species phylogenetic relatedness (Appendix S2). Both
approaches indicated that species recruitment and density-
dependent parameters displayed large deviations from a Brown-
ian model of evolution and are consistent with an interpretation
of trait evolution under an Ornstein–Uhlenbeck (OU) process
(Appendix S2). In addition, Blomberg’s K-values across all
species also indicated a significant phylogenetic signal according
to an OU process (KRP =0.249, P=0.006; KDD =0.087,
P=0.038; KCDD =0.147, P=0.051; lower Blomberg’s K-values
under an OU model indicate a stronger phylogenetic signal).
This phylogenetic signal was partially driven by differences
between angiosperm and gymnosperm species. The RP of gym-
nosperms was significantly higher (F1,162 =4.23, P=0.048) and
their DD marginally weaker (F1,162 =2.67, P=0.108), although
both taxonomic groups showed a similar CDD (F1,162 =1.31,
P=0.247). In summary, all methods of analysis indicated that
species average RP, DD and CDD are phylogenetically con-
served, so they were considered a species characteristic and were
subsequently mapped using species distributions to test whether
community geographical patterns can be predicted by climatic
variables and species functional traits. Note that in any case we
aim to test the relative contribution of phylogeny to the geo-
graphical patterns of RP, DD and CDD and to their functional
predictors.
Climatic and species trait information
We selected 20 variables to examine how climate and species
functional traits are associated with RP, DD and CDD. We
selected four climatic variables representing the two most
important climatic gradients occurring at meso- and macro-
scales in the eastern USA (temperature and precipitation). Two
measures of temperature (BIO5, maximum temperature of the
warmest month; BIO6, minimum temperature of the coldest
month) and two measures of precipitation (BIO12, annual pre-
cipitation; BIO18, precipitation in the warmest month) were
extracted from the 30 arcsec WorldClim database (http://
www.worldclim.org). Sixteen further variables representing a
range of important axes of ecological strategies of tree species
(Westoby et al., 2002; Westoby & Wright, 2006; Moles et al.,
2007; Kraft et al., 2008) were obtained from various sources.
This trait information is available at the species level (Table 1,
Appendix S3). These traits show geographical variation at local
scales as well as large scales, and none of them were strongly
correlated (r<0.8). We did not include the widely used plant
trait of specific leaf area (SLA) in the analysis due to limited
information (values were only available for fewer than 50% of
the species).
Forest community data and geographical analyses
with Random Forest models
To generate community data at the site level from species-level
data we first created a presence–absence matrix for the 164
species across FIA sites. Then we calculated averages of the three
dependent variables (RP, DD, CDD) and the 16 functional trait
predictors at sites including at least two species.
We then generated Random Forest models, each based on 200
regression trees with 1000 permutations per regression tree, in
the R package RandomForest (Liaw & Wiener, 2002) –to statis-
tically account for (1) the spatial patterns of average community
data across sites and (2) the variation across species without the
spatial component. A comparison of both analyses served to test
whether the drivers of recruitment match between community
and species levels. Briefly, Random Forest modelling is a pow-
erful machine-learning technique that combines the predictions
of multiple independent regression trees into a robust compo-
site model. The relative importance of the predictors is assessed
by the decrease in explained variance resulting from permuta-
tions of the focal variable. We selected Random Forest modelling
over more traditional linear modelling approaches because it
does not assume stationarity of relationships. This statistical
technique is able to disentangle interacting effects and can iden-
tify nonlinear and scale-dependent relationships that often
occur at the scale of this analysis among multiple, correlated
predictors (Cutler et al., 2007).
Forest recruitment and species traits
Global Ecology and Biogeography,24, 192–202, © 2014 John Wiley & Sons Ltd 195
For the community-level analysis we generated nine Random
Forest models, proceeding as follows. For each dependent vari-
able (RP, DD and CDD) we generated three models considering
the average site values of the predictors based on the subplot
species composition: one including only climatic variables, one
including only species traits and one including all predictors
(Appendix S4). This allows us to evaluate the extent to which the
climate versus trait predictors with higher importance values
have independent relationships with the geographical structure
of the response variables or if their relative importance is largely
due to correlations with the other class of predictors. For each
model we recorded the percentage of the explained variance
(pseudo-R2ranging from 0 to 100%) and ranked the relative
importance of each variable ranging from 100 (the strongest
predictor) to 0 (having no predictive power) according to its
node purity value. The sign of the relationship between the
dependent variable and the predictors was further assessed with
Pearson correlation.We found that models including all predic-
tors explained large amounts of variance, which might be due to
our inclusion of 20 predictors. We therefore regenerated the
Random Forest models including only those predictors with
high relative importance. To take a conservative approach the
threshold was set to an importance value of over 75. These
simplified models explained 12–20% less variance than the full
models (Appendix S4), but in all cases remained powerful in
terms of explained variance considering the small size of the FIA
plots, the large sample size and the geographical scope of the
data.
For the species-level analysis we generated three Random
Forest models, one for each dependent variable, considering
trait values for each species as the predictors rather than site
averages. Here, we also recorded the percentage of explained
variance of the model, and we estimated the relative importance
of each predictor and the sign of the relationship between the
dependent variable and the predictors as indicated above.
In the final step of the analysis we generated spatial auto-
correlograms of Moran’s I(the most widely used autocor-
relation metric) to evaluate the strengths of the geographical
gradients of RP, DD and CDD, followed by correlograms of the
residuals extracted from the Random Forest model combining
all predictors. Moran’s Ivalues, ranging from c.1 to 1, indicate
whether communities connected at a given distance are more
similar (positive correlation) or less similar (negative correla-
tion) than expected for randomly associated pairs of plots. This
allows us to evaluate the extent to which the climatic variables
and traits are able to account for the spatial patterns across the
full range of scales, from local to subcontinental (Hawkins et al.,
2014).
RESULTS
Recruitment potential (density-independent effects)
The average RP of local communities has a strong geographical
pattern very similar to the distribution of the minimum
number of frost-free days (Min F-F days; Table 1) required by
each species to complete its seasonal phenological develop-
ment, and to a lesser extent to seed size (Figs 1a, 2a & 3a,b).
Local communities comprising species able to live under
longer seasonal freezing conditions and with smaller seeds had
the potential to recruit three times more individuals (Table 2).
Geographically, the potential for recruitment of tree commu-
nities in the forests of the eastern USA has a latitudinal
gradient, with higher recruitment at northern latitudes, an
elevational gradient with higher recruitment in Appalachian
forests compared with surrounding lowland sites (Fig. 1a) and
a visible longitudinal gradient with lowland forests nearer the
east coast having higher recruitment than forests at the forest–
prairie interface.
Density dependence
Most species suffer reduced recruitment in the presence of
neighbours (i.e. a negative DD) (Appendix S5); thus negative
density dependence in tree communities is expected to occur
across most, but not all, of the region (Fig. 1b, inset). The geo-
graphical structure of DD was relatively weaker than for RP,
with which it was also weakly correlated (r=0.17) (see Fig. 1b
for a map and spatial correlogram; DD showed a flatter curve
for the Moran’s I-values of the raw data than RP). The spatial
structure of DD at intermediate scales resulted, for instance, in
stronger density-dependent effects at forest–prairie interface
sites near the Texas–Louisiana–Arkansas borders associated
with oak–pine forest (Fig. 1b). Nevertheless, we found two
clear trait predictors: communities comprising species with
higher wood densities and trunk diameters tend to interact
more negatively (Table 2, Figs 2b & 3d). In addition, tree
communities also show a sharp shift from negative to
positive DD associated with longer, more severe winters
(Figs 1b & 3e).
Conspecific density dependence
Average CDD shows similarities with DD. Firstly, their magni-
tudes were comparable. Second,negative CDD is also prevalent.
Third, the geographical pattern of CDD is uncorrelated with RP
(r=−0.02), at least partially because species with stronger nega-
tive CDD interactions occur in numerous scattered locations at
smaller geographical scales, such as the lower basin of the Mis-
sissippi River, parts of Florida, lowland forest along the east
coast, Appalachian forest and Great Lakes forest.Fourth, there is
again a shift from negative to more positive conspecific interac-
tions moving northward and westward related to Min F-F days
(Figs 1c & 3a), although the spatial pattern is weak, albeit still
significant (P<0.001), based on the correlogram. A major dif-
ference is that leaf N instead of wood density was the primary
trait predictor of CDD (Table 2), with stands having lower leaf
N showing the potential to interact more negatively between
conspecifics (positive correlation, Fig. 2c). Another difference is
that CDD was predicted by a larger number of traits than DD,
including seed mass, maximum trunk diameter, wood density
and anaerobic tolerance (Table 2, Fig. 2c).
O. Godoy et al.
Global Ecology and Biogeography,24, 192–202, © 2014 John Wiley & Sons Ltd196
Species traits and climate: independent and
correlated predictors
It is worth noting that the combined Random Forest models
including both climate and species traits as predictors explained
large amounts of the observed geographical patterns of RP, DD
and CDD (Fig. 2) and captured virtually all the spatial pattern;
the Moran’s I-values of the model residuals for each variable
retained no detectable spatial autocorrelation at any scale
(Fig. 1). In addition, Random Forest models at the community
level explained more variation of RP, DD and CDD than models
at the species level, perhaps because of the significant differences
in sample size (cf. Table 2, Fig. 2). Surprisingly, climatic vari-
ables per se are not strong predictors; the pseudo-R2did not
increase appreciably between the Random Forest community
model including only the species traits and the model including
both climate and traits (Fig. 2, Appendix S4). However, the tem-
perature variables, maximum temperature during summer
(BIO5) (for RP) and minimum temperature during winter
(BIO6) (for CDD), were related to spatial patterns via their
relationships with specific species traits (BIO5 and BIO6 are
correlated with Min F-F days, r=0.56 and r=0.62, respectively;
Fig. 3c,e). Moreover, the geographical patterns for density-
independent (RP) and density-dependent (DD and CDD) pro-
cesses were not identical, the former being much more strongly
structured than the latter. This suggests that the geographical
structure of the magnitude and sign of community density-
dependent processes does not depend on the recruitment poten-
tial although they are modestly correlated at the species level
(rRP–DD =−0.40, rRP–CDD =−0.22).
DISCUSSION
Our central finding is that the recruitment of trees species from
density-independent and density-dependent processes in
eastern US forest communities exhibits broad-scales patterns.
We also found clear functional predictors. Among them, vari-
ation in the magnitude of recruitment potential has the clearest
geographical structure, explained by the ability to tolerate
seasonal freezing conditions (Min F-F days), seed mass and
Figure 1 Left column: (a) geographical
pattern of mean recruitment potential
(RP), (b) mean strength of density
dependence (DD), and (c) mean strength
of conspecific density dependence (CDD)
for the eastern forests of the USA. The
inset in (b) shows change from negative
to positive interactions moving
northwards. Right column: spatial
correlograms of RP, DD and CDD using
random samples of 15,000 sites. Raw data
and residuals correspond to Random
Forest models provided in Fig. 2. All
Moran’s I-values for the residuals are
between 0.02 and 0.03.
Forest recruitment and species traits
Global Ecology and Biogeography,24, 192–202, © 2014 John Wiley & Sons Ltd 197
maximum temperature during summer (Table 2, Figs 1a & 2).
Several explanations for why forest communities at northern
latitudes, at higher elevations and along the east coast can recruit
up to three times more individuals are plausible and not mutu-
ally exclusive. First, species have evolved smaller seeds in their
adaptation to cold climates in order to complete rapid seed
development and avoid freeze-induced mortality in the autumn
(Moles et al., 2007; Hawkins et al., 2014). Because species with
smaller seeds tend to produce more offspring per individual
than species with larger seeds, we expect that being more fecund
may generate higher community recruitment since the seed pre-
dation rate is generally not related to seed mass (Moles et al.,
2003). Second, the activity of several drivers of seed/seedling
mortality such as pathogens, insects and mammals increases
with temperature, mainly under moist conditions (Bale et al.,
2002; Harvell et al., 2002). Thus, this biotic risk of mortality is
reduced by being adapted to tolerate freezing conditions for
longer when active growth is not possible (i.e. low Min F-F
days). Finally, a longer growing season with a hot summer
increases seedling mortality rates resulting from seasonal
drought and water stress (Ruiz-Benito et al., 2013).
We also found that density-dependent processes, either in the
interactions with the total number of adults (DD) or with
conspecific adults only (CDD), reduced recruitment at almost
all spatial scales (Fig. 1b,c). Two distinctive functional traits arise
when explaining the magnitude of negative interactions: wood
density for DD and leaf nitrogen for CDD (Table 2, Fig. 2b,c).A
simple explanation for DD that is partially supported by our
analyses may be that tree species with higher wood density tend
to be larger (Table 2, Max. trunk diameter) (Baker et al., 2004),
and bigger individuals produce stronger negative competitive
effects on both conspecific and heterospecific neighbours (Rees,
2013). Why communities with lower leaf N only tend to interact
more negatively between conspecifics can be related to shade
tolerance (Kobe & Vriesendorp, 2011), although our categorical
measure of shade tolerance was unable to account for the geo-
graphical pattern of CDD. Shade-tolerant species also show
enhanced survival against attack by host-specific enemies
because they invest in conservative functional strategies such as
low leaf N, large seeds and dense wood (traits with high relative
importance in our analyses; Table 2) (Kitajima, 1994; Coley &
Barone, 1996; Hewitt, 1998). Also, lower leaf N can reflect
Figure 2 Variable importance values
from a Random Forest model (based on
200 regression trees) of (a) mean of
recruitment potential (RP), (b) mean of
the strength of density dependence (DD),
and (c) mean of the strength of
conspecific density dependence (CDD),
including all trait and environmental
predictors. Broad-scale environmental
variables are in white and traits are in
gray. Pearson’s correlations are given for
those important predictors to provide the
sign of the relationship. Non-abbreviated
value names are given in Table 1. Similar
models performed only with the most
important predictors are presented in
Appendix S4.
O. Godoy et al.
Global Ecology and Biogeography,24, 192–202, © 2014 John Wiley & Sons Ltd198
species living on nutrient-poor soils (Fig. 3f; lower Leaf N values
correspond to communities living in ultisols, which are acid
forest soils with relatively low fertility). Competition for nutri-
ents in these environments is probably high compared with in
more fertile soils, especially with conspecifics as they exploit
similar soil resources.
Tree communities shift from negative to positive DD and
CDD from south to north, which relates to the latitudinal
pattern of Min F-F days (Figs 1b,c & 3a). This change in the sign
of species interactions (at c. 100 Min F-F days) with increasing
environmental stress (colder conditions at northern latitudes) is
consistent with studies at the global scale supporting the stress-
gradient hypothesis (Callaway et al., 2002). Seedlings are the life
stage most sensitive to cold injury because the layer of cold air
close to the ground can freeze their meristems; an effect that is
worsened in forest openings by radiation frost (Howe et al.,
2003). Thus, the presence of adult trees is likely to favour seed-
ling survival at northern latitudes by increasing the soil surface
temperature. Nevertheless, the fact that the change in the sign of
DD and CDD is abrupt and the other main predictors are func-
tional traits raises doubts about a simple climate-driven expla-
nation. It may be possible that negative to positive interactions
depend partially on the evolutionary histories of plants. This is
a plausible explanation given that we found DD marginally
weaker for gymnosperm species, and changes in composition
from angiosperm-dominated forests (oak–hickory, aspen–
birch) to gymnosperm-dominated forests (spruce–fir) (Appen-
dix S6) appear to match the line separating negative from
positive interactions (most evident again for DD).
One surprising result from both the trait and the combined
(climate +trait) Random Forest community models is that Min
F-F days is consistently the predictor having the highest relative
Figure 3 Geographical pattern of the
mean of the most important predictors
resulting from the Random Forest
analyses. For a detailed description see
Table 1.
Forest recruitment and species traits
Global Ecology and Biogeography,24, 192–202, © 2014 John Wiley & Sons Ltd 199
importance (Fig. 2, Appendix S4). It is commonly thought that
the ability to respond to seasonally freezing temperatures defines
range limits. At northern latitudes freezing temperatures limit
fruit maturation, and at southern latitudes budburst may not
occur due to the lack of chilling temperatures. Indeed, the
physiological relationship of freezing to species occurrences has
been extensively explored using process-based species distribu-
tion models to project shifts in the species ranges due to climate
change (reviewed in Chuine, 2010). Our results suggest an
aspect related to demography that has not been widely consid-
ered. Specifically, within species ranges the degree of adaptation
to the seasonal breadth of freezing (and thus the length of the
growing season) influences the magnitude of potential recruit-
ment and the strength and sign of the density dependence of
local communities. We believe that a combination of previous
work with a consideration of the role of phenology could stimu-
late research to predict not only future forest composition with
environmental change (Cleland et al., 2007) but also the relative
abundances of tree species.
In sum, by combining demographic, functional and geo-
graphical aspects of local forest communities, we have found
strong broad-scale patterns in three key components of forest
recruitment. We have also found that a phenological trait related
to the tolerance to seasonal freezing conditions combined with
three functional traits (seed mass, wood density and leaf N)
account for most of the observed spatial structure. Importantly,
we have quantified the magnitude by which these traits explain
reduction in recruitment or the shift from negative to positive
interactions. Nevertheless, we estimated an average value of RP,
DD and CDD per species, and the observed geographical vari-
ation was based on differences in species occurrences among
local communities. Future research needs to estimate these three
parameters within species at smaller spatial scales to address the
question of which locally operating drivers can change the
species potential for recruitment and the strength of density
dependence, and hence local community patterns. This may
account for the 12, 25 and 40% of variation in RP, DD and CDD,
respectively, that is unexplained by our models [there is no
residual spatial autocorrelation after fitting the predictors
(Fig. 1), indicating that no additional spatially structured vari-
ables are needed to explain the patterns statistically]. Coupled
with this, we also need more data on intraspecific trait variation,
at least across sites (such information is logistically difficult to
obtain, but it could initially be focused on seed mass, wood
density and leaf N based on the results found here). Finally, the
subplots sampled by the FIA are small (0.017 hectares), which
means that much of the unexplained variance could be due to
sampling error with respect to the trees actually present in local
communities, a potential problem that is difficult to access at the
subcontinental scale. These methodological issues aside, our
results show that forest recruitment exhibits broad-scale pat-
terns and can be explained by a few key functional traits.
ACKNOWLEDGEMENTS
We thank Lorena Gómez-Aparicio, Irena Šímová, Pedro Villar,
Jeff Diez and Elena Granda for their comments. O.G. and M.R.
acknowledge post-doctoral financial support provided by the
Spanish Ministry of Education and Science and Fulbright Com-
mission (FU-2009-0039 and BVA-2010-0596, respectively).
REFERENCES
Adler, P.B., Salguero-Gómez, R., Compagnoni, A., Hsu, J.S.,
Ray-Mukherjee, J., Mbeau-Ache, C. & Franco, M. (2014)
Functional traits explain variation in plant life history strat-
egies. Proceedings of the National Academy of Sciences USA,
111, 740–745.
Baker, T.R., Phillips, O.L., Malhi, Y., Almeida, S., Arroyo, L.,
Di Fiore, A., Erwin, T., Killeen, T.J., Laurance, S.G. &
Laurance, W.F. (2004) Variation in wood density determines
spatial patterns in Amazonian forest biomass. Global Change
Biology,10, 545–562.
Bale, J.S., Masters, G.J., Hodkinson, I.D., Awmack, C., Bezemer,
T.M., Brown, V.K., Butterfield, J., Buse, A., Coulson, J.C. &
Farrar, J. (2002) Herbivory in global climate change research:
direct effects of rising temperature on insect herbivores.
Global Change Biology,8, 1–16.
Bolker, B.M. (2008) Ecological models and data in R. Princeton
University Press, Princeton, NJ.
Bolker, B.M. & Pacala, S.W. (1999) Spatial moment equations
for plant competition: understanding spatial strategies and
the advantages of short dispersal. The American Naturalist,
153, 575–602.
Borchert, M.I., Davis, F.W., Michaelsen, J. & Oyler, L.D. (1989)
Interactions of factors affecting seedling recruitment of blue
oak (Quercus douglasii) in California. Ecology,70, 389–404.
Table 2 Random Forest models at the species level including all
trait predictors (environmental predictors are excluded). Shown
are variable importance values (VI) of the species traits with VI
>75 for recruitment potential, density dependence and
conspecific density dependence. A value of 100 indicates the
species trait with the highest relative importance for the Random
Forest model. The sign of Pearson’s correlation are given to
indicate the sign of the relationship. The percentage of explained
variance (Pseudo-R2) gives an overall fit for the model.
Recruitment
potential
(RP)
Density
dependence
(DD)
Conspecific
density
dependence
(CDD)
Pseudo-R20.31 0.28 0.22
Species trait
Anaerobic tolerance 80.7 ()
Leaf N mass 100 ()
Max. trunk diameter 95.4 () 75.1 ()
Min. frost-free days 87.6 () 81.0 () 85.3 ()
Seed mass 100 () 75.7 ()
Wood density 100 () 78.3 ()
O. Godoy et al.
Global Ecology and Biogeography,24, 192–202, © 2014 John Wiley & Sons Ltd200
Callaway, R.M., Brooker, R.W., Choler, P., Kikvidze, Z., Lortie,
C.J., Michalet, R., Paolini, L., Pugnaire, F.I., Newingham, B.,
Aschehoug, E.T., Armas, C., Kikodze, D. & Cook, B.J. (2002)
Positive interactions among alpine plants increase with stress.
Nature,417, 844–848.
Carnicer, J., Coll, M., Pons, X., Ninyerola, M., Vayreda, J. &
Peñuelas, J. (2014) Large-scale recruitment limitation in
Mediterranean pines: the role of Quercus ilex and forest suc-
cessional advance as key regional drivers. Global Ecology and
Biogeography,23, 371–384.
Chave, J., Coomes, D., Jansen, S., Lewis, S.L., Swenson, N.G. &
Zanne, A.E. (2009) Towards a worldwide wood economics
spectrum. Ecology Letters,12, 351–366.
Chesson, P. (2000) Mechanisms of maintenance of species
diversity. Annual Review of Ecology and Systematics,31, 343–
366.
Chuine, I. (2010) Why does phenology drive species distribu-
tion? Philosophical Transactions of the Royal Society B: Biologi-
cal Sciences,365, 3149–3160.
Clark, J.S., Macklin, E. & Wood, L. (1998) Stages and spatial
scales of recruitment limitation in southern Appalachian
forests. Ecological Monographs,68, 213–235.
Clark, J.S., Bell, D., Chu, C., Courbaud, B., Dietze, M., Hersh, M.,
HilleRisLambers, J., Ibáñez, I., LaDeau, S. & McMahon, S.
(2010) High-dimensional coexistence based on individual
variation: a synthesis of evidence. Ecological Monographs,80,
569–608.
Cleland, E.E., Chuine, I., Menzel, A., Mooney, H.A. & Schwartz,
M.D. (2007) Shifting plant phenology in response to global
change. Trends in Ecology and Evolution,22, 357–365.
Coley, P.D. & Barone, J. (1996) Herbivory and plant defenses in
tropical forests. Annual Review of Ecology and Systematics,27,
305–335.
Coll, M., Peñuelas, J., Ninyerola, M., Pons, X. & Carnicer, J.
(2013) Multivariate effect gradients driving forest demo-
graphic responses in the Iberian Peninsula. Forest Ecology and
Management,303, 195–209.
Comita, L.S., Muller-Landau, H.C., Aguilar, S. & Hubbell, S.P.
(2010) Asymmetric density dependence shapes species
abundances in a tropical tree community. Science,329, 330–
332.
Connell, J.H. (1971) On the role of natural enemies in pre-
venting competitive exclusion in some marine animals
and in rain forest trees. Dynamics of populations (ed. by
J. Den Boer and G. Gradwell), pp. 298–312. PUDOC,
Wageningen.
Cutler, D.R., Edwards, T.C. Jr, Beard, K.H., Cutler, A., Hess, K.T.,
Gibson, J. & Lawler, J.J. (2007) Random forests for classifica-
tion in ecology. Ecology,88, 2783–2792.
Diniz-Filho, F., Alexandre, J., Rangel, T.F., Santos, T. &
Mauricio Bini, L. (2012) Exploring patterns of interspecific
variation in quantitative traits using sequential phylogenetic
eigenvector regressions. Evolution,66, 1079–1090.
Enquist, B.J., West, G.B., Charnov, E.L. & Brown, J.H. (1999)
Allometric scaling of production and life-history variation in
vascular plants. Nature,401, 907–911.
Harms, K.E., Wright, S.J., Calderon, O., Hernandez, A. & Herre,
E.A. (2000) Pervasive density-dependent recruitment
enhances seedling diversity in a tropical forest. Nature,404,
493–495.
Harvell, C.D., Mitchell, C.E., Ward, J.R., Altizer, S., Dobson, A.P.,
Ostfeld, R.S. & Samuel, M.D. (2002) Climate warming and
disease risks for terrestrial and marine biota. Science,296,
2158–2162.
Hawkins, B.A., Rueda, M., Rangel, T.F., Field, R. & Diniz-Filho,
J.A.F. (2014) Community phylogenetics at the biogeographi-
cal scale: cold tolerance, niche conservatism and the structure
of North American forests. Journal of Biogeography,41, 23–38.
Herrera, C.M., Jordano, P., Guitián, J. & Traveset, A. (1998)
Annual variability in seed production by woody plants and
the masting concept: reassessment of principles and relation-
ship to pollination and seed dispersal. The American Natural-
ist,152, 576–594.
Hewitt, N. (1998) Seed size and shade-tolerance: a comparative
analysis of North American temperate trees. Oecologia,114,
432–440.
Howe, G.T., Aitken, S.N., Neale, D.B., Jermstad, K.D., Wheeler,
N.C. & Chen, T.H. (2003) From genotype to phenotype:
unraveling the complexities of cold adaptation in forest trees.
Canadian Journal of Botany,81, 1247–1266.
Janzen, D.H. (1970) Herbivores and the number of tree species
in tropical forests. The American Naturalist,104, 501–528.
Johnson, D.J., Beaulieu, W.T., Bever, J.D. & Clay, K. (2012)
Conspecific negative density dependence and forest diversity.
Science,336, 904–907.
Kitajima, K. (1994) Relative importance of photosynthetic traits
and allocation patterns as correlates of seedling shade toler-
ance of 13 tropical trees. Oecologia,98, 419–428.
Kobe, R.K. & Vriesendorp, C.F. (2011) Conspecific density
dependence in seedlings varies with species shade tolerance in
a wet tropical forest. Ecology Letters,14, 503–510.
Kraft, N.J., Valencia, R. & Ackerly, D.D. (2008) Functional traits
and niche-based tree community assembly in an Amazonian
forest. Science,322, 580–582.
Lambers, J.H.R., Clark, J.S. & Beckage, B. (2002) Density-
dependent mortality and the latitudinal gradient in species
diversity. Nature,417, 732–735.
Levin, S.A. (1992) The problem of pattern and scale in ecology:
the Robert H. MacArthur award lecture. Ecology,73, 1943–
1967.
Liaw, A. & Wiener, M. (2002) Classification and regression by
RandomForest. RNews,2, 18–22.
McIntire, E.J. & Fajardo, A. (2011) Facilitation within species: a
possible origin of group-selected superorganisms. The Ameri-
can Naturalist,178, 88–97.
Moles, A.T., Warton, D.I. & Westoby, M. (2003) Do small-seeded
species have higher survival through seed predation than
large-seeded species? Ecology,84, 3148–3161.
Moles, A.T., Ackerly, D.D., Tweddle, J.C., Dickie, J.B., Smith, R.,
Leishman, M.R., Mayfield, M.M., Pitman, A., Wood, J.T. &
Westoby, M. (2007) Global patterns in seed size. Global
Ecology and Biogeography,16, 109–116.
Forest recruitment and species traits
Global Ecology and Biogeography,24, 192–202, © 2014 John Wiley & Sons Ltd 201
Mueller, R.C., Scudder, C.M., Porter, M.E., Talbot Trotter, R.,
Gehring, C.A. & Whitham, T.G. (2005) Differential tree mor-
tality in response to severe drought: evidence for long-term
vegetation shifts. Journal of Ecology,93, 1085–1093.
Niinemets, Ü. (1997) Role of foliar nitrogen in light harvesting
and shade tolerance of four temperate deciduous woody
species. Functional Ecology,11, 518–531.
Peñuelas, J. & Boada, M. (2003) A global change-induced biome
shift in the Montseny mountains (NE Spain). Global Change
Biology,9, 131–140.
Rees, M. (2013) Competition on productivity gradients – what
do we expect? Ecology Letters,16, 291–298.
Rees, M., Condit, R., Crawley, M., Pacala, S. & Tilman, D. (2001)
Long-term studies of vegetation dynamics. Science,293, 650–
655.
Rejmánek, M. & Richardson,D.M. (1996) What attributes make
some plant species more invasive? Ecology,77, 1655–1661.
Ribbens, E., Silander, J.A. Jr & Pacala, S.W. (1994) Seedling
recruitment in forests: calibrating models to predict patterns
of tree seedling dispersion. Ecology,75, 1794–1806.
Ruiz-Benito, P., Lines, E.R., Gómez-Aparicio, L., Zavala, M.A. &
Coomes, D.A. (2013) Patterns and drivers of tree mortality in
Iberian forests: climatic effects are modified by competition.
PLoS ONE,8, e56843.
Ruiz-Benito, P., Gómez-Aparicio, L., Paquette, A., Messier, C.,
Kattge, J. & Zavala, M.A. (2014) Diversity increases carbon
storage and tree productivity in Spanish forests. Global
Ecology and Biogeography,23, 311–322.
Smith, C.C. & Fretwell, S.D. (1974) The optimal balance
between size and number of offspring. The American Natural-
ist,108, 499–506.
Sterck, F., Markesteijn, L., Schieving, F. & Poorter, L. (2011)
Functional traits determine trade-offs and niches in a tropical
forest community. Proceedings of the National Academy of Sci-
ences USA,108, 20627–20632.
Swenson, N.G. & Enquist, B.J. (2007) Ecological and evolution-
ary determinants of a key plant functional trait: wood density
and its community-wide variation across latitude and eleva-
tion. American Journal of Botany,94, 451–459.
Westoby, M. & Wright, I.J. (2006) Land-plant ecology on the
basis of functional traits. Trends in Ecology and Evolution,21,
261–268.
Westoby, M., Falster, D.S., Moles, A.T., Vesk, P.A. & Wright, I.J.
(2002) Plant ecological strategies: some leading dimensions of
variation between species. Annual Review of Ecology and Sys-
tematics,33, 125–159.
Zhu, K., Woodall, C.W. & Clark, J.S. (2012) Failure to migrate:
lack of tree range expansion in response to climate change.
Global Change Biology,18, 1042–1052.
SUPPORTING INFORMATION
Additional supporting information may be found in the online
version of this article at the publisher’s web-site.
Appendix S1 Forest Inventory and Analysis plot design and
examples of the model.
Appendix S2 Phylogenetic comparative analyses.
Appendix S3 Sources selected for trait information.
Appendix S4 Additional Random Forest models.
Appendix S5 Species values of recruitment potential and
density dependence.
Appendix S6 Geographical pattern of angiosperm and gymno-
sperm forest communities.
BIOSKETCHES
Oscar Godoy is a post-doctoral fellow interested in
the functional and evolutionary determinants of species
interactions, mainly plant competition, and their
consequences for community and ecosystem dynamics.
Marta Rueda is a post-doctoral fellow interested in
biogeography and global ecological and evolutionary
patterns. Her current research interest also includes
fragmentation theory and community phylogenetics at
broad scales.
Bradford A. Hawkins is interested in ecological and
phylogenetic patterns across a range of spatial scales,
with a focus on linking local and biogeographical
processes.
Author contributions: O.G., M.R. and B.A.H. conceived
the ideas, prepared the data, performed the analyses and
wrote the manuscript.
Editor: Ian Wright
O. Godoy et al.
Global Ecology and Biogeography,24, 192–202, © 2014 John Wiley & Sons Ltd202
... Specifically, tree performance at upper elevational edges is strongly related to cold temperatures and precipitation, while lower elevational limits may be determined by a combination of climate and regeneration failure, and competition fails to explain growth and performance declines in adult trees at both distributional limits (Ettinger et al. 2011, Ettinger and HilleRisLambers 2013, Copenhaver-Parry and Cannon 2016). Regeneration failure, however, could be influenced by the high sensitivity of seedlings to competition at lower distribution edges, which is consistent with evidence indicating that regeneration patterns of North American trees may be strongly influenced by biotic interactions (Dobrowski et al. 2015, Godoy et al. 2015. Species distributions are typically only represented by the occurrence of adult individuals, and regeneration dynamics of North American trees across broad spatial scales have been little evaluated (but see Bell et al. 2013); these findings highlight the need for further investigation into the links between biotic interactions, regeneration dynamics, and distribution patterns and suggest that seedling distributions may reveal relationships between distributions and biotic interactions that may be masked when only adult individuals are considered (see Life stage). ...
... In eastern North American forests, tree seedlings exhibit a trend of increasingly strong negative density dependence toward their southern distributional limits, and this trend is strongly correlated with recruitment potential (Godoy et al. 2015). Similarly, eastern North American seedlings exhibit strong positive density dependence at northern distributional limits, indicating that facilitation may be an important mechanism promoting seedling establishment in more stressful climates. ...
... Similarly, eastern North American seedlings exhibit strong positive density dependence at northern distributional limits, indicating that facilitation may be an important mechanism promoting seedling establishment in more stressful climates. Yet, while biotic interactions certainly appear to contribute to the distribution patterns of eastern North American tree seedlings, climate sensitivity, particularly to seasonal freezing temperatures, still appears to be the dominant factor explaining seedling recruitment patterns (Godoy et al. 2015). Furthermore, seedlings can show evidence of local adaptation to combinations of temperature and moisture availability (Eickmeier et al. 1975), which can enable new climatic conditions to be suitable to a given species generally but not to the extant populations of seedlings in a given area. ...
Article
Full-text available
Species distributions have often been assumed to represent climatic limitations, yet recent evidence has challenged these assumptions and emphasized the potential importance of biotic interactions, dispersal limitation, and disturbance. Despite significant investigation into these factors, an integrated understanding of where and when they may be important is lacking. Here, we review evidence for the factors underlying the historical and contemporary distributions of North American tree species and argue that a cohesive conceptual framework must be informed by an understanding of species ecological and evolutionary history. We further demonstrate that available evidence offers little indication of a significant, independent influence of biotic interactions or dispersal limitation on species distributions. Disturbance may provide important constraints on distributions in limited contexts. Overall, historic and contemporary evidence suggests that species distributions are strongly influenced by climate, yet examples of disequilibrium with climate abound. We propose that differences among life stages and the impacts of human land use may contribute to explain these inconsistencies and are deserving of greater research attention.
... can also influence the strength and impact of CNDD (Brown et al., 2020;Godoy et al., 2015;Lebrija-Trejos et al., 2016;McCarthy-Neumann & Ibáñez, 2013;Pu et al., 2020). Hardy and Sonké (2004) suggested that dispersal limitation might be the main factor influencing conspecific clustering (e.g., short-distance dispersal species only fail to arrive at all suitable sites), leading to variation in conspecific neighbor densities among species (Wright, 2002). ...
... Our results that those species with lower LDMC suffered stronger CNDD can well support this view. Moreover, McCarthy-Neumann and Ibáñez (2013) and Godoy et al. (2015) found a positive relationship between leaf nitrogen levels and susceptibility of seedling to CNDD, and shadetolerant species tended to obtain lower level of leaves nitrogen content. Thus, shade-tolerant tree species are considered to be better defense against the attack from pathogens or other enemies (Kobe & Vriesendorp, 2011;McCarthy-Neumann & Ibáñez, 2013;McCarthy-Neumann & Kobe, 2008 which helped them to have stronger resistance to diseases, pathogens, and competition in the light-limited understory environment when they were young (Tinya & Ódor, 2016). ...
Article
Full-text available
en Conspecific negative density dependence (CNDD) is an important mechanism facilitating species coexistence in plant communities. Yet, which factors influence CNDD is still not clear. Here, we examined the factors influencing CNDD in tree species within a 60‐ha tropical rainforest dynamics plot. Point pattern analysis was used to test the existence and strength of CNDD. The relationships between strength of CNDD and plant functional traits, as well as abiotic factors, were assessed by stepwise regression analysis. Results showed that 89% of the examined species exhibited CNDD, and the strength of CNDD decreased with increasing spatial scales and species abundance. The maximum strength of CNDD (dmax) was negatively related to leaf dry matter content and positively related to potential maximum height. Soil water content, available nitrogen, and total potassium were the major abiotic factors determining dmax. Our study suggests that CNDD is prevalent in most tree species and that both abiotic (soil) and biotic (species abundance, plant functional traits) factors constrain the strength of CNDD in our tropical montane rainforest. Abstract in Chinese is available with online material. 抽象 fr 摘要:负密度制约效应是促进植物群落物种共存的重要机制。但是,究竟什么因素可以影响负密度制约强度并不清楚。因此,本研究在热带山地雨林60公顷大样地内检验了何种因素可以影响负密度制约强度。通过点格局方法检验了负密度制约效应及其强度,并通过多元逐步回归评估了功能性状及非生物因素对负密度制约效应强度的影响。结果表明样地内89%的物种均表现出负密度制约效应且该效应随空间尺度的增加和多度的增大逐渐减少。此外,负密度制约效应还和叶片干物质含量负相关,并与物种的潜在最大高度正相关。土壤含水量、土壤速效氮含量以及土壤全钾含量是影响负密度制约效应的主要土壤因素。该研究表明,负密度制约效应在热带山地雨林普遍存在且非生物因素(土壤因素)及生物因素(功能性状、物种多度)均能够对负密度制约强度起到限制性效应。
... One issue arising in biogeographic tests of biotic resistance and propagule pressure is that natural recruitment often covaries with latitude. For example, forest communities at higher latitudes of eastern North America recruit up to three times more trees than at lower latitudes (Godoy, Rueda, & Hawkins, 2015). Marine invertebrates also exhibit a seasonal, large recruitment pulse in high latitudes, whereas low latitudes produce a lower year-round recruitment signal (Connolly, Menge, & Roughgarden, 2001;Freestone, Osman, & Whitlatch, 2009). ...
... All Didemnum recruits perished and only algae and one limpet were found (at lower right) averaged across the year, but this remains to be tested. In addition to species interaction strength, seasonality also influences patterns of recruitment, further complicating colonization dynamics (Connolly et al., 2001;Godoy et al., 2015). Given that high latitude sites exhibit greater temporal variation in environmental regime and propagule pressure than tropical sites, we expect that high latitude sites would be most sensitive to the timing of experimentation and sampling. ...
Article
Full-text available
Aim Recent work has documented latitudinal gradients of biotic resistance, revealing diminished invasion success in the tropics as compared to the temperate zone. However, no studies have explored the biogeography of biotic resistance simultaneously with propagule pressure, which can greatly influence invasion dynamics and covary with latitude. Location 9–41° latitude, north‐western Atlantic seagrass beds. Methods We conducted field experiments to test the interactive effects of propagule pressure (experimentally placed recruits) and biotic resistance (predation) on invader performance in temperate and tropical seagrass beds. For these experiments, we used marine invertebrate propagules from bryozoans (Bugula neritina) and tunicates (Didemnum spp.). We also quantified natural recruitment with and without exposure to predators. Results Surprisingly, predation substantially reduced invader survival at almost all latitudes. Overall, invaders experienced 15%–27% survival with predation as opposed to 75%–87% survival without predation. These patterns did not change when we increased local scale propagule pressure of Bugula by over 2‐fold. However, predation had no effect on invader survival in Florida, where natural recruitment was up to 500‐fold greater than other sites. We also measured substantial in situ recruitment of Bugula onto bare experimental surfaces that was not diminished with exposure to predators at mid‐latitudes, suggesting a regional scale predator swamping effect. Conclusions Contrary to recent findings of latitudinal variation in biotic resistance, we found that predation strongly reduced invader success in both temperate and tropical seagrass beds. However, our results also indicate that propagule pressure (natural recruitment) can influence invasion at the regional scale to overwhelm native communities. Our data suggest that predation and propagule pressure act at varying spatial scales to affect biogeographic patterns of invasion. The importance of latitudinal variation in these interactions is largely untested but deserves attention given that globalization will continue to facilitate opportunities for invasion.
... This incorporation of the multidimensional nature of the phenotype-fitness relationship has recently been successfully demonstrated within habitats for forest and grassland species (e.g. Godoy et al. 2015;R€ uger et al. 2018;Breitschwerdt et al. 2019;P erez-Ramos et al. 2019). ...
Article
Full-text available
Life history strategies are fundamental to the ecology and evolution of organisms and are important for understanding extinction risk and responses to global change. Using global datasets and a multiple response modelling framework we show that trait‐climate interactions are associated with life history strategies for a diverse range of plant species at the global scale. Our modelling framework informs our understanding of trade‐offs and positive correlations between elements of life history after accounting for environmental context and evolutionary and trait‐based constraints. Interactions between plant traits and climatic context were needed to explain variation in age at maturity, distribution of mortality across the lifespan and generation times of species. Mean age at maturity and the distribution of mortality across plants’ lifespan were under evolutionary constraints. These findings provide empirical support for the theoretical expectation that climatic context is key to understanding trait to life history relationships globally. Life history strategies are fundamental to the ecology and evolution of organisms and are important for understanding extinction risk and responses to global change. Using global datasets and a multiple response modelling framework we show that the relationship between plant traits and life history strategies depend on climatic and evolutionary contexts, for a diverse range of species at the global scale.
... Therefore, depending on the nature of the interactions, competition and predation can affect the importance of each other (Chase et al., 2002), with the predominant interaction in a given scenario promoting or limiting diversity (Chesson & Kuang, 2008). This scenario becomes even more complex, considering that propagule pressure can modulate the effects of competition and predation (Cheng et al., 2019), and may also vary biogeographically (Cheng et al., 2019;Connolly, Menge, & Roughgarden, 2001;Godoy, Rueda, & Hawkins, 2015). ...
Article
Full-text available
Aims The importance of biotic interactions in creating and maintaining diversity is expected to increase towards low latitudes. However, the way in which predation affects diversity can depend on how predators mediate competitive interactions and also on defensive traits of prey. Here, we assessed the role of physical defences of prey to escape predation and how the importance of predation on community structure and diversity changes across latitude. Location Six sites, in three regions distributed across 45 degrees of latitude in the Atlantic Ocean: a tropical region in Angola, a subtropical region in Brazil and a temperate region in Wales, UK. Methods We manipulated predation on marine sessile communities, using exclusion cages and assessed community parameters, including their susceptibility to biological invasion during early and advanced succession. Results Predation was more intense in the tropics and in advanced communities suggesting that predation effects increase through time. In the tropical region, predators reduced the number of co‐occurring species and beta diversity, limited the occurrence of exotic species and promoted a change in the identity of the dominant organisms, replacing soft‐bodied organisms with calcified animals. In the subtropical region, predation promoted a similar trait‐mediated change in the identity of dominant prey, although it was not strong enough to affect diversity and did not prevent bioinvasion. In the temperate region, other processes than predation seem to drive the community organization and resistance to invasion. Main conclusions Our results support both Biotic Interaction and Biotic Resistance Hypotheses, showing that the importance of predation to biodiversity increases towards the tropics. In addition, where predation is intense, morphological traits of prey drive the final structure and dominance in the community. Our results suggest that physical defences are the main traits preventing predation, perhaps explaining why calcified organisms are among the most common invasive species in coastal habitats.
... A generally appreciated alternative to tackle this complex network of plant-plant and plant-environment interactions uses species' traits rather than relying on their taxonomical identity 4,5 . There is growing evidence of the relevance of functional traits for determining the three main components of individual performance-growth, reproduction, and survival [6][7][8] , as well as the strength and sign of plant interactions [9][10][11] . However, it is still poorly understood how interspecific trait differences connect with the stabilizing and equalizing mechanisms that jointly determine biodiversity maintenance 12,13 . ...
Article
Full-text available
Functional traits are expected to modulate plant competitive dynamics. However, how traits and their plasticity in response to contrasting environments connect with the mechanisms determining species coexistence remains poorly understood. Here, we couple field experiments under two contrasting climatic conditions to a plant population model describing competitive dynamics between 10 annual plant species in order to evaluate how 19 functional traits, covering physiological, morphological and reproductive characteristics, are associated with species' niche and fitness differences. We find a rich diversity of univariate and multi-dimensional associations, which highlight the primary role of traits related to water-and light-use-efficiency for modulating the determinants of competitive outcomes. Importantly, such traits and their plasticity promote species coexistence across climatic conditions by enhancing stabilizing niche differences and by generating competitive trade-offs between species. Our study represents a significant advance showing how leading dimensions of plant function connect to the mechanisms determining the maintenance of biodiversity.
... A generally appreciated alternative to tackle this complex network of plant-plant and plant-environment interactions uses species' traits rather than relying on their taxonomical identity 4,5 . There is growing evidence of the relevance of functional traits for determining the three main components of individual performance-growth, reproduction, and survival [6][7][8] , as well as the strength and sign of plant interactions [9][10][11] . However, it is still poorly understood how interspecific trait differences connect with the stabilizing and equalizing mechanisms that jointly determine biodiversity maintenance 12,13 . ...
Preprint
Full-text available
Functional traits are expected to modulate plant competitive dynamics. However, how traits and their plasticity in response to contrasting environments connect with the mechanisms determining species coexistence remains poorly understood. Here, we couple field experiments under two contrasting climatic conditions to a plant population model describing competitive dynamics between 10 annual plant species in order to evaluate how 19 functional traits, covering physiological, morphological and reproductive characteristics, are associated with species’ niche and fitness differences. We find a rich diversity of univariate and multidimensional associations, which highlight the primary role of traits related to water- and light use-efficiency for modulating the determinants of competitive outcomes. Importantly, such traits and their plasticity promote species coexistence across climatic conditions by enhancing stabilizing niche differences and by generating competitive trade-offs between species. Our study represents a significant advance showing how leading dimensions of plant function connect to the mechanisms determining the maintenance of biodiversity
... Tolerance to abiotic stresses, mainly extreme temperatures, water and light availability, is a key driver of plant distributions and competitive interactions (Valladares and Niinemets 2008, Martinez-Tilleria et al. 2012, Godoy et al. 2015, Díaz et al. 2016, Kunstler et al. 2016. As a consequence of the episodic declines in global temperatures after the Eocene Thermal Optimum (ca 55-45 Ma), freezing winter temperatures probably represent the most severe environmental stress with which plants growing in the northern temperate zones have had to contend (Latham and Ricklefs 1993, Graham 1999, Donoghue 2008. ...
Article
Adaptive syndromes and their evolutionary constraints represent a powerful construct for understanding plant distributions. However, it is unclear how the species requirements to face multiple stressors promotes syndrome formation and to which abiotic stressors these syndromes show adaptive value over broad geographic scales. We combined local occurrence data from the U.S. Forest Inventory and Analysis (FIA) of 219 angiosperm and 85 gymnosperm species living across the conterminous US with phylogenies and trait data to identify tree syndromes, their evolutionary conservatism, and their adaptive value over broad scales. Factor analyses and evolutionary model selection revealed that trees possess functional trait syndromes that are strongly conserved. Major syndromes at the species level differed between angiosperms and gymnosperms. While the two main syndromes in angiosperms were related to cold and drought-waterlogging tolerance, in gymnosperms a trade-off between shade and drought tolerance was the main syndrome followed by a growth-fire resistance syndrome. Additional RLQ and fourth-corner approaches revealed that trait syndromes at the community level were broadly similar to those observed at the species level for angiosperms, although this was less clear for gymnosperms. This suggests that syndrome evolution has played an important role on angiosperm distributions, whereas additional ecological factors explain gymnosperm distributions. Importantly, syndromes show adaptive value, as they were geographically associated with several environmental variables showing structure from continental to local scales, being temperature the main abiotic stressor. Our results indicate that across the conterminous US tree species possess clear syndromes that are subjected to strong evolutionary constraints driving tree species and forest community distribution. This article is protected by copyright. All rights reserved.
Article
Climate and competition interact to affect species’ performance, such as growth and survival, and help determine species distributions and coexistence. However, it is unclear how climatic conditions modulate frequency‐dependent performance ‐ i.e. how performance changes as a species becomes locally rare or common. This is critical because declines in performance as a species becomes more common (negative frequency‐dependence), is a signature of niche differences among species that stabilizes coexistence, whereas positive frequency‐dependence leads to priority effects and hampers species coexistence. Here, we used dendrochronology and hierarchical models to test whether frequency‐dependent growth of sugar pine (Pinus lambertiana) depends on climatic conditions. We found that growth rates were strongly dependent on annual precipitation, but no frequency‐dependence was evident across all years. However, there was a strong interaction between precipitation and frequency‐dependence, revealing stabilizing niche differences in dry years but positive frequency‐dependence in wet years. These differences emerged due to precipitation‐driven changes in the direction and strength of both con‐ and heterospecific competition. Overall, these results show how stabilizing and destabilizing effects can be temporally dynamic for long‐lived species and interact with climate variation.
Article
1.The maintenance of tree diversity has been explained by multiple mechanisms. One of the most thoroughly studied is conspecific negative density dependence, in which specialist plant enemies reduce survivorship of seeds, seedlings, or saplings located near adult conspecifics. Although there is much support that conspecific negative density dependence occurs in temperate forests, only a subset of the species investigated thus far exhibit this recruitment pattern. It remains unclear what drives differential susceptibility to conspecifics among tree species. Previous investigators have considered shade tolerance and mycorrhizal type (arbuscular mycorrhizal vs. ectomycorrhizal association) as two traits that might explain differential susceptibility to conspecific negative density dependence. 2. Here, we test whether these two plant traits predict susceptibility of tree saplings to conspecific negative density dependence in a temperate hardwood forest using three responses: spatial point patterns of saplings, sapling growth, and sapling survival. 3. Spatial patterns of saplings indicate that shade tolerant species are less sensitive to conspecifics than shade intolerant species, but show no differences based on mycorrhizal type. Conversely, shade tolerant saplings exhibit reduced growth, but not survival, when located in areas with high conspecific density. We interpret this finding in light of the conservative functional strategies of shade tolerant species, which typically have low leaf nitrogen levels and slower growth to divert resources to tissue defence against enemies. We found a marginally significant effect of mycorrhizal type interacting with adult conspecific density, where arbuscular mycorrhizal species show a greater reduction in growth than ectomycorrhizal species in areas dense with conspecifics. 4. Synthesis: We conclude that the shade tolerance level and the mycorrhizal type of temperate forest saplings may influence how their growth and survival respond to the adult conspecific trees in their neighbourhoods. This article is protected by copyright. All rights reserved.
Article
Full-text available
An important aim of plant ecology is to identify leading dimensions of ecological variation among species and to understand the basis for them. Dimensions that can readily be measured would be especially useful, because they might offer a path towards improved worldwide synthesis across the thousands of field experiments and ecophysiological studies that use just a few species each. Four dimensions are reviewed here. The leaf mass per area-leaf lifespan (LMA-LL) dimension expresses slow turnover of plant parts (at high LMA and long LL), long nutrient residence times, and slow response to favorable growth conditions. The seed mass-seed output (SM-SO) dimension is an important predictor of dispersal to establishment opportunities (seed output) and of establishment success in the face of hazards (seed mass). The LMA-LL and SM-SO dimensions are each underpinned by a single, comprehensible tradeoff, and their consequences are fairly well understood. The leaf size-twig size (LS-TS) spectrum has obvious consequences for the texture of canopies, but the costs and benefits of large versus small leaf and twig size are poorly understood. The height dimension has universally been seen as ecologically important and included in ecological strategy schemes. Nevertheless, height includes several tradeoffs and adaptive elements, which ideally should be treated separately. Each of these four dimensions varies at the scales of climate zones and of site types within landscapes. This variation can be interpreted as adaptation to the physical environment. Each dimension also varies widely among coexisting species. Most likely this within-site variation arises because the ecological opportunities for each species depend strongly on which other species are present, in other words, because the set of species at a site is a stable mixture of strategies.
Article
Full-text available
The fossil record has led to a historical explanation for forest diversity gradients within the cool parts of the Northern Hemisphere, founded on a limited ability of woody angiosperm clades to adapt to mid-Tertiary cooling. We tested four predictions of how this should be manifested in the phylogenetic structure of 91,340 communities: (1) forests to the north should comprise species from younger clades (families) than forests to the south; (2) average cold tolerance at a local site should be associated with the mean family age (MFA) of species; (3) minimum temperature should account for MFA better than alternative environmental variables; and (4) traits associated with survival in cold climates should evolve under a niche conservatism constraint.
Article
Full-text available
Ecologists seek general explanations for the dramatic variation in species abundances in space and time. An increasingly popular solution is to predict species distributions, dynamics, and responses to environmental change based on easily measured anatomical and morphological traits. Trait-based approaches assume that simple functional traits influence fitness and life history evolution, but rigorous tests of this assumption are lacking, because they require quantitative information about the full lifecycles of many species representing different life histories. Here, we link a global traits database with empirical matrix population models for 222 species and report strong relationships between functional traits and plant life histories. Species with large seeds, long-lived leaves, or dense wood have slow life histories, with mean fitness (i.e., population growth rates) more strongly influenced by survival than by growth or fecundity, compared with fast life history species with small seeds, short-lived leaves, or soft wood. In contrast to measures of demographic contributions to fitness based on whole lifecycles, analyses focused on raw demographic rates may underestimate the strength of association between traits and mean fitness. Our results help establish the physiological basis for plant life history evolution and show the potential for trait-based approaches in population dynamics.
Article
A plant lineage can compete for resources in a spatially variable environment by colonizing new areas, exploiting resources in those areas quickly before other plants arrive to compete with it, or tolerating competition once other plants do arrive. These specializations are ubiquitous in plant communities, but all three have never been derived from a spatial model of community dynamics—instead, the possibility of rapid exploitation has been either overlooked or confounded with colonization. We use moment equations, equations for the mean densities and spatial covariance of competing plant populations, to characterize these strategies in a fully spatial stochastic model. The moment equations predict endogenous spatial pattern formation and the efficacy of spatial strategies under different conditions. The model shows that specializations for colonization, exploitation, and tolerance are all possible, and these are the only possible spatial strategies; among them, they partition all of the endogenous spatial structure in the environment. The model predicts two distinct short‐dispersal specializations where parents disperse their offspring locally, either to exploit empty patches quickly or to fill patches to exclude competitors.
Article
AimLarge-scale patterns of limitations in tree recruitment remain poorly described in the Mediterranean Basin, and this information is required to assess the impacts of global warming on forests. Here, we unveil the existence of opposite trends of recruitment limitation between the dominant genera Quercus and Pinus on a large scale and identify the key ecological drivers of these diverging trends. LocationSpain Methods We gathered data from the Spanish National Forest inventory to assess recruitment trends for the dominant species (Pinus halepensis, Pinus pinea, Pinus pinaster, Pinus nigra, Pinus sylvestris, Pinus uncinata, Quercus suber, Quercus ilex, Quercus petraea, Quercus robur, Quercus faginea and Quercus pyrenaica). We assessed the direct and indirect drivers of recruitment by applying Bayesian structural equation modelling techniques. ResultsSevere limitations in recruitment were observed across extensive areas for all Pinus species studied, with recruitment failure affecting 54-71% of the surveyed plots. In striking contrast, Quercus species expanded into 41% of the plots surveyed compared to only 10% for Pinus and had a lower local recruitment failure (29% of Quercus localities compared to 63% for Pinus species). Bayesian structural equation models highlighted the key role of the presence of Q.ilex saplings and the increase in the basal area of Q.ilex in limiting recruitment in five Pinus species. The recruitment of P.sylvestris and P.nigra showed the most negative trends and was negatively associated with the impacts of fire. Main conclusionsThis study identified Q.ilex, the most widespread species in this area, as a key driver of recruitment shifts on a large scale, negatively affecting most pine species with the advance of forest succession. These results highlight that the future expansion/contraction of Q.ilex stands with ongoing climate change will be a key process indirectly controlling the demographic responses of Pinus species in the Mediterranean Basin.
Article
It is argued that the problem of pattern and scale is the central problem in ecology, unifying population biology and ecosystems science, and marrying basic and applied ecology. Applied challenges, such as the prediction of the ecological causes and consequences of global climate change, require the interfacing of phenomena that occur on very different scales of space, time, and ecological organization. Furthermore, there is no single natural scale at which ecological phenomena should be studied; systems generally show characteristic variability on a range of spatial, temporal, and organizational scales. The observer imposes a perceptual bias, a filter through which the system is viewed. This has fundamental evolutionary significance, since every organism is an "observer" of the environment, and life history adaptations such as dispersal and dormancy alter the perceptual scales of the species, and the observed variability. It likewise has fundamental significance for our own study of ecological systems, since the patterns that are unique to any range of scales will have unique causes and biological consequences. The key to prediction and understanding lies in the elucidation of mechanisms underlying observed patterns. Typically, these mechanisms operate at different scales than those on which the patterns are observed; in some cases, the patterns must be understood as emerging form the collective behaviors of large ensembles of smaller scale units. In other cases, the pattern is imposed by larger scale constraints. Examination of such phenomena requires the study of how pattern and variability change with the scale of description, and the development of laws for simplification, aggregation, and scaling. Examples are given from the marine and terrestrial literatures.
Article
Biodiversity loss could reduce primary productivity and the carbon storage provided by forests; however, the mechanisms underpinning the effects of biodiversity on multiple ecosystem functions are not completely understood. Spanish forests are of particular interest because of the broad variation in environmental conditions and management history. We tested for the existence of a relationship between diversity effects and both carbon storage and tree productivity, and examined the relative importance of complementarity and selection mechanisms in a wide variety of forests, from cold deciduous Atlantic to xeric Mediterranean evergreen forests.
Article
The relationship between the energy expended per offspring, fitness of offspring, and parental fitness is presented in a two-dimensional graphical model. The validity of the model in determining an optimal parental strategy is demonstrated analytically. The model applies under various conditions of parental care and sibling care for the offspring but is most useful for species that produce numerous small offspring which are given no parental care.
Article
A precise knowledge of forest demographic gradients in the Mediterranean area is essential to assess future impacts of climate change and extreme drought events. Here we studied the geographical patterns of forest demography variables (tree recruitment, growth and mortality) of the main species in Spain and assessed their multiple ecological drivers (climate, topography, soil, forest stand attributes and tree-specific traits) as well as the geographical variability of their effects and interactions. Quantile modeling analyses allowed a synthetic description of the gradients of multiple covariates influencing forest demography in this area. These multivariate effect gradients showed significantly stronger interactions at the extremes of the rainfall gradient. Remarkably, in all demographic variables, qualitatively different levels of effects and interactions were observed across tree-size classes. In addition, significant differences in demographic responses and effect gradients were also evident between the dominant genus Quercus and Pinus. Quercus species presented significantly higher percentage of plots colonized by new recruits, whereas in Pinus recruitment limitation was significantly higher. Contrasting positive and negative growth responses to temperature were also observed in Quercus and Pinus, respectively. Overall, our results synthesize forest demographic responses across climatic gradients in Spain, and unveil the interactions between driving factors operating in the drier and wetter edges.