demographic drivers of successional changes in …swensonn/nordenetal_2012_ecology.pdfecology, 93(8)...

Ecology, 93(8) Supplement, 2012, pp. S70–S82� 2012 by the Ecological Society of America

Demographic drivers of successional changes in phylogeneticstructure across life-history stages in plant communities

NATALIA NORDEN,1,2,5 SUSAN G. LETCHER,3 VANESSA BOUKILI,2 NATHAN G. SWENSON,4 AND ROBIN CHAZDON2

1Departamento de Ciencias Biologicas, Universidad de los Andes, Carrera 1 numero 18A-10, Bogota, D.C., Colombia2Department of Ecology and Evolutionary Biology, University of Connecticut, Storrs, Connecticut 06269-3043 USA

3Organization for Tropical Studies, Apartado Postal 676 2050, San Pedro de Montes de Oca, Costa Rica4Department of Plant Biology, Michigan State University, East Lansing, Michigan 48824 USA

Abstract. To gain insight into the ecological processes driving community reassembly indisturbed ecosystems, we assessed the phylogenetic dispersion of early- and late-successionaltree species occurring in lowland forests of northeastern Costa Rica. Early-successional specieswere more closely related than expected by chance, whereas late-successional species tended tobe less closely related than expected by chance. Then, we evaluated temporal changes in thephylogenetic structure of seedling and tree assemblages in four 1-ha plots of secondary forestsin this region. We found an increase in the phylogenetic evenness among tree individuals overtime in all secondary tree assemblages, indicating that relatedness among tree individualsdecreases as succession unfolds. This pattern was jointly promoted by recruitment andmortality processes, suggesting that increasing evenness was caused by the replacement ofindividuals of early-successional species from closely related lineages by late-successionalspecies belonging to a wider diversity of lineages. Based on species occurrence, however, treecommunity reassembly did not show any significant phylogenetic trend over time. Theseresults suggest that shifts in species abundance over succession have a greater impact on thephylogenetic structure of the community than the turnover of species. Seedling assemblagesshowed higher phylogenetic evenness than tree assemblages, suggesting that propagulecolonization is an important process driving phylogenetic changes in species compositionthroughout succession. Overall, our findings showed that the phylogenetic structure of thesesuccessional communities varies at two temporal scales. At short timescales, decreaseddominance by early-successional species over succession leads to increased evenness amongtree individuals. At longer timescales, colonization processes result in increased phylogeneticevenness in seedling communities compared to tree communities, forecasting increasingphylogenetic evenness among adult individuals at late-successional stages.

Key words: community assembly; Costa Rica; life-history stages of plants; mortality; phylogeneticstructure; recruitment; succession; vegetation dynamics.

INTRODUCTION

Disturbed ecosystems are increasing in extent across

the landscape, to the degree that successional forest

cover exceeds mature forest cover in many regions

(ITTO 2002, FAO 2005). Despite the growing role of

these successional communities in the conservation of

biodiversity, climate change mitigation, and many

ecosystem services (Chazdon et al. 2009), recent

theoretical developments in plant ecology have focused

on species-rich, mature plant communities, rather than

on less diverse secondary forests (Webb 2000, Hubbell

2001, Muller-Landau et al. 2006, Jabot and Chave 2009,

Kraft and Ackerley 2010, Swenson et al. 2012).

Secondary forests, however, provide an ideal laboratory

for testing ecological theories, as they reflect community

assembly in action. So far, successional models have

been based on the premise that species turnover through

time is dictated by differences in growth rates and life

span (Finegan 1996, Rees et al. 2001). These attributes

define species’ ecological syndromes (Westoby et al.

2002) and underpin most explanations of temporal

changes in the community during secondary succession

(Rees et al. 2001). Yet, these theoretical approaches do

not consider the phylogenetic background of co-

occurring species. Linking observed patterns of commu-

nity structure to processes occurring over evolutionary

timescales offers a compelling framework for the

development of a synthetic ecological theory (Webb et

al. 2002, Cavender-Bares et al. 2009, Vamosi et al. 2009).

Moreover, it provides valuable information for the

assessment of species conservation status and anthro-

pogenic impacts (Warwick and Clarke 1998, Helmus et

al. 2010).

Manuscript received 13 November 2010; revised 22 March2011; accepted 4 April 2011. Corresponding Editor: J.Cavender-Bares. For reprints of this Special Issue, seefootnote 1, p. S1.

5 Present address: Departamento de Ecologıa y Territorio,Pontificia Universidad Javeriana, Transversal 4 number 42-00, Bogota, D.C., Colombia.E-mail: [email protected]

S70

Recent studies have shown that disturbed ecosystems

contain more closely related species than non-perturbed

ones (Verdu and Pausas 2007, Knapp et al. 2008,

Helmus et al. 2010, Cavender-Bares and Reich 2012),

suggesting that species’ response to disturbance has a

phylogenetic component. If so, the phylogenetic relat-

edness exhibited by early-successional species should

translate into a temporal change in the phylogenetic

structure of the community as succession unfolds. Yet

uncertainty persists regarding this issue, since the few

studies evaluating the effect of disturbance on the

relatedness of co-occurring species are limited to the

comparison of perturbed and non-perturbed communi-

ties (Verdu and Pausas 2007, Knapp et al. 2008, Helmus

et al. 2010). These single-time investigations represent a

snapshot of species composition in time and space,

ignoring the fact that succession is a continuous and

dynamic process. Integrating a temporal dimension into

studies of phylogenetic community structure should add

critical elements to disentangle the different mechanisms

driving community reassembly after disturbance.

In the most comprehensive study addressing this issue,

Letcher (2010) evaluated changes in the phylogenetic

structure of successional tropical forests in a replicated

chronosequence of 30 sites in northeastern Costa Rica.

By dividing tree plots into five stand age categories, she

found increased phylogenetic evenness (i.e., overdisper-

sion sensu Webb et al. 2002) over succession. Further,

Letcher (2010) found that small stems were less closely

related than large stems, suggesting that the increased

phylogenetic evenness observed over succession was

related to the recruitment of late-successional species.

These findings concur with previous studies (Verdu and

Pausas 2007, Knapp et al. 2008, Helmus et al. 2010), and

suggest that communities at the onset of succession are

mainly comprised of closely related species, whereas late

stages of succession contain species belonging to a wide

array of lineages. Indeed, pioneer species in the

neotropics typically belong to the genus Cecropia

(Urticaceae) or to the family Malvaceae sensu lato

(both groups belonging to the Rosid clade). Contrast-

ingly, since the most common environmental condition

in tropical forests’ understory is shade, functional

convergence on traits giving adaptive value for regener-

ation under low light availability (Hubbell 2005) would

lead to phylogenetic evenness among late-successional

species. So far, the degree of relatedness among early- or

late-successional species has never been evaluated

quantitatively, partly because classifying species into

these ecological syndromes is a challenging task (Wright

et al. 2003).

In this study, we assessed the phylogenetic relatedness

of early- and late-successional species, and tracked

temporal changes in phylogenetic community structure

within secondary forest plots and between seedling and

tree stages, in a lowland Neotropical rainforest. By

integrating the current knowledge in successional

ecology (Finegan 1996, Chazdon 2008) and prior

information from studies evaluating community phylo-

genetic structure in disturbed systems (Verdu andPausas 2007, Knapp et al. 2008, Helmus et al. 2010,

Letcher 2010), we propose a likely scenario for thetemporal changes in the phylogenetic structure of

successional communities. Assuming that early-succes-sional species are clustered within the angiospermphylogeny and that late-successional species show a

more even phylogenetic distribution, we expect phylo-genetic evenness to increase over succession. Such a

pattern can be the result of two simultaneous processes:the recruitment of distantly related late-successional

species/individuals, and the death of closely relatedearly-successional species/individuals. As part of the

same process, small size classes, which represent futuretree recruits, and thus reflect the species composition of

more advanced successional stages (Norden et al. 2009),should also show higher phylogenetic evenness than the

standing tree community. How recruitment and mortal-ity affect the phylogenetic structure of the community

not only depends upon relatedness among species/individuals entering or leaving the community, but also

on the relatedness among these groups and the group ofsurvivors. For instance, we expect increasing evennessover succession to be partially associated with the

recruitment of distantly related species/individuals.However, increased evenness can also be the outcome

of the recruitment of species/individuals that are closelyrelated to one another, but distantly related to the pool

of survivors.We tested these different scenarios using robust

successional data on seedling and tree dynamics fromlowland forest plots in northeastern Costa Rica (see

Plate 1). Specifically, we quantified the phylogeneticrelatedness of early- and late-successional species using

static data from several studies conducted in bothsecond- and old-growth forests in this area. This

information provided us with an appropriate frameworkto interpret temporal changes in the phylogenetic

structure of secondary forest communities. We usedlong-term data on recruitment and mortality of trees (12

years) and seedlings (5 years) in four secondary-forestpermanent plots in this same region to assess changes inphylogenetic structure over time. We compared the

temporal trends in tree assemblages with those observedin mature forests to evaluate whether phylogenetic

structure in secondary forests approaches that of matureforest as succession unfolds. Then, we evaluated how

recruitment and mortality contribute to the temporalchanges observed in the phylogenetic structure of the

plant community over succession.

MATERIALS AND METHODS

Study site

The study was conducted at La Selva BiologicalStation and surrounding areas of northeastern Costa

Rica. This region is classified as tropical lowland wetforest, with an annual precipitation of ;3900 mm and

August 2012 S71PHYLOGENETIC STRUCTURE OVER SUCCESSION

an average annual temperature of 26.58C (McDade et al.

1994). The landscape matrix is composed of old-growth

and second-growth forest patches surrounded by agri-

cultural lands, pastures, and cash crops. Site elevations

range from 40 to 200 m above sea level. Land-use

history was similar within all secondary sites.

Phylogenetic dispersion in ecological syndromes

We tested the assumptions that early-successional

species are phylogenetically clustered within the regional

flora and that late-successional species are phylogeneti-

cally even by evaluating the phylogenetic dispersion in a

categorical variable defining species as second-growth

specialists or old-growth specialists. To do so, we assessed

whether early-successional species tended to be more

closely related than expected by chance, and late-

successional species more distantly related than expected

by chance. Because the pioneer vs. non-pioneer catego-

rization is seen as false dichotomy (Wright et al. 2003), we

defined a third category, the generalists, for which we did

not have any particular expectation regarding its

phylogenetic dispersion. To assign species to one of these

three categories, we used data on species abundance in

mature and secondary stands throughout the study area

from various sources (Chazdon et al. 2011). This data set

was composed of trees � 10 cm diameter at breast height

(dbh) sampled in 18.3 ha of old-growth forest and 11.3 ha

of secondary forest. Secondary forests ranged from 5 to

45 years since abandonment and included forests

regenerating on abandoned pastures and in cleared areas

that were never grazed. Second-growth and old-growth

sites were interspersed across a broad landscape, thereby

avoiding the pitfalls of confounding habitat differences

with spatial variables (Chazdon et al. 2011). A previous

study using part of the data used here showed that

patterns of floristic similarity among tree plots were

independent of spatial distance (Norden et al. 2009).

Thus, the potential biasing effect of spatial autocorrela-

tion seems to be minimal.

Our statistical approach classified each species into

one of each of the three categories based on its

abundance in secondary and mature stands (Chazdon

et al. 2011). If a species is a second-growth specialist,

then we would expect that pi, the relative abundance of

this species in second-growth forests, is sufficiently

higher than pi, the relative abundance in this species in

old-growth forests; or equivalently, the ratio pi/( piþ pi )is sufficiently higher than a constant k. Using a

multinomial model, this statistical test procedure deter-

mines for each species whether the observed data

provide sufficient evidence that their ratio is statistically

higher than this pre-specified constant k. Here, k was

designated as 0.5, a liberal threshold based on a ‘‘simple

majority’’ rule, which allowed us to classify most species

into one of these categories (Chazdon et al. 2011). Old-

growth forest specialists are similarly classified as species

with a statistically higher ratio of individuals in old-

growth than in second-growth forest stands, and

generalists are classified as those that have statistically

indistinguishable relative abundance in old-growth and

second-growth stands.

Among the 359 species occurring in these plots, 203

could be assigned with certainty to one of these

categories using this statistical procedure, and the

remaining species were too rare to classify. We

constructed the phylogenetic supertree of these 203

species using the database Phylomatic (Webb and

Donoghue 2005). Phylomatic starts with the family-

level tree of angiosperms and considers genera as

polytomies within families and species as polytomies

within genera. The latest Angiosperm Phylogeny

Group (APG) classification (Phylomatic tree version

R20080417; P. F. Stevens, available online)6 was used for

the supertree backbone. The BLADJ algorithm was

implemented in the community phylogenetic software

Phylocom (v. 4.0.b; Webb et al. 2008) to calibrate the

species pool supertree by applying known molecular and

fossil dates (Wikstrom et al. 2001) to nodes in the

supertree, and to reduce variance among branch lengths

by evenly spacing nodes of unknown ages. The same

procedure was used in the subsequent analyses.

The degree of phylogenetic dispersion in each of the

three ecological syndromes defined was assessed by

measuring the mean phylogenetic distance (MPD)

among the 58 species classified as second-growth

specialists, among the 107 species classified as old-

growth specialists, and among the 38 species classified as

generalists. We assessed significance for each category

by comparing the observed value of MPDs to the null

distribution of MPD values obtained by shuffling the

species names across the tips of the phylogeny 999 times

(Chazdon et al. 2003). Early-successional species were

considered significantly clustered within the regional

species pool of 203 species if the observed MPD of the

58 second-growth specialists was in the lower 5% of the

null distribution (one-tailed test). Likewise, late-succes-

sional species were considered significantly even within

the regional species pool if the observed MPD of the 107

old-growth specialists was in the higher 5% of the null

distribution. Because we did not have any particular

expectation for generalists, these species were considered

significantly clustered or even within the regional species

pool if the observed MPD of the 38 generalists was in

the lower or higher 2.5% of the null distribution (two-

tailed test). Although there are numerous analytical

methods to measure phylogenetic dispersion in a trait

(Blomberg et al. 2003, Webb et al. 2008, Fritz and Purvis

2010), we believe this method to be the most appropriate

considering the presence of polytomies in our tree, and

the fact that we are analyzing a categorical variable.

Calculations were performed with the R statistical

software (v. 2.11.1; R Development Core Team 2010).

6 http://www.mobot.org/MOBOT/research/APweb/

NATALIA NORDEN ET AL.S72 Ecology Special Issue

Temporal changes in phylogenetic community structure

To evaluate temporal changes in phylogenetic struc-

ture within sites and between life stages, we used long-

term data on seedling and tree dynamics. Permanent

plots of 1 ha were established in 1997 in four secondary

forest sites, and two additional 1-ha plots were

established in 2005 within mature forest patches (Table

1). In each plot, all trees with dbh of �5 cm were tagged,

mapped, and measured for diameter at 1.3 m height.

Thereafter, censuses were completed annually. In each

census, new recruits into the �5 cm diameter class were

tagged, mapped, and measured; and stems were

recorded as dead if no living tissues could be observed.

The final tree census of all plots was completed in 2008.

In each of the secondary forest sites, 144 seedling plots

of 1 3 5 m were created in 1998 (Capers et al. 2005). In

each plot, all woody stems �20 cm in height and ,1 cm

dbh were tagged and identified. Subsequently, seedling

censuses were conducted every 8 months. During each

census, newly recruited seedlings were tagged and

identified, and missing seedlings were recorded as dead.

Individuals that moved into the next size class (.1 cm

dbh) during the census period were considered as

survivors. The final seedling census was completed in

August 2003 (Capers et al. 2005). We did not include

seedling data from mature forest plots since these data

are not available. Trees and seedlings were identified by

experienced field assistants, and vouchers were collected

to compare with botanical specimens at the La Selva

Herbarium and the Instituto Nacional de Biodiversidad

in Costa Rica (INBio). Although seedlings of all life

forms were censused, here we consider only the canopy

and subcanopy trees and palms, as well as treelet species

reaching the .5 cm dbh size class, to be consistent with

tree censuses.

We constructed an appropriate regional species pool

including all the canopy tree, palm, and treelet species

that reach dbh �5 cm and that could potentially colonize

the plots based on a database of vascular plants occurring

in the Sarapiquı region (Proyecto Flora Digital de La

Selva 2009). We excluded non-angiosperm taxa because

they are rare in the region and could contribute

disproportionally to phylogenetic structure owing to their

low relatedness with most of the species. The final list for

the community species pool included a total of 603

species. The total number of species used for this

phylogenetic supertree was higher than the one used in

the previous analysis, which only included species for

which we had information about their life-history status.

A phylogenetic tree for those species was constructed

using Phylomatic (Webb and Donoghue 2005) according

to the methods described in the section Phylogenetic

dispersion of ecological syndromes. Because the Phylo-

matic algorithm maps unknown relationships as polyto-

mies on the supertree, measures of phylogenetic structure

using these phylogenies may have potential biases and

loss of statistical power (Swenson 2009). Letcher (2010)

tested for this issue using a very similar species regional

pool to ours, and found that the phylogenetic measures

calculated using the Phylomatic tree were highly corre-

lated to those obtained using a tree where all families

were collapsed into polytomies. Indeed, loss of phyloge-

netic resolution generally causes false negative results,

rather than false positives (Swenson 2009), which would

underestimate the strength of our results.

We evaluated temporal changes in the phylogenetic

structure of tree and seedling assemblages in each plot

by calculating the net relatedness index (NRI; Webb

2000) using the package ‘‘picante’’ (Kembel et al. 2010)

in the R statistical software (v. 2.11.1; R Development

Core Team 2010). Similar results were obtained using

the nearest taxon index, and are not reported here. Over

the timescale of our study, floristic changes during

succession might reflect changes in species’ relative

abundance more than the gain and/or loss of species.

We thus performed this analysis using both abundance-

and incidence-based data. NRI for each local assem-

TABLE 1. Site descriptions of 1-ha monitoring plots in northeastern Costa Rica.

Plot name andabbreviation Location

Age in 1997(yr) Previous land use Landscape matrix

Lindero Sur, LSUR La Selva BiologicalStation

12 cattle pasture for 5–6 yr adjacent to large tractof old-growth forest

Tirimbina, TIR Finca de ArturoSalazar, adjacentto TirimbinaResearch Center

15 cattle pasture for .6 years;may have been cut twice

surrounded by matrixof pasture andsecond-growth forest

Lindero El Peje, LEPS La Selva BiologicalStation

20 cattle pasture for 5–6 yr adjacent to large tractof old-growth forest

Cuatro Rios, CR La Virgen deSarapiquı

25 cleared for cattle pasture,little or no grazing(,3 yr)

surrounded by matrixof pasture andsecond-growth forest

Selva Verde, SV Chilamate mature forest no history of logging, butarea has been subject tohunting

surrounded by pastureand second-growthforest

LEP primary, LEPP La Selva BiologicalStation

mature forest no history of logging,protected from hunting

surrounded by old-growth forest onthree sides


blage was calculated using the following formula (Webb

et al. 2002):

NRI ¼ �MPD�MPDrnd

sdMPDrnd

where MPD is the mean phylogenetic distance of co-

occurring taxa in local assemblages, MPDrnd is the mean

and sdMPDrnd the standard deviation of 999 MPDs

obtained from randomly generated local assemblages.

We used the Phylogeny Tip Shuffle null model in

Phylocom, which shuffles species labels across the entire

phylogeny. Although permuting elements within the

community data matrix is a more conservative method

for avoiding Type I error (Hardy 2008), randomizing

relatedness among species and/or individuals is a more

appropriate null model in temporal analyses. Random-

izing elements within the community data matrix

changes the identity of surviving stems between consec-

utive censuses, inflating the temporal turnover in the null

model, and biasing the observed result toward a lower

than expected turnover. Randomizing species labels

across the phylogeny avoids this problem and has the

additional benefit of maintaining the observed occupan-

cy rates, the plot abundance distribution and plot

species richness. Thus, if individuals within a species

are dispersal limited or have spatial contagion in their

observed distribution, this observed pattern will be

maintained in all null communities generated. Alterna-

tive null models, such as a swap algorithm, maintain

occupancy rates, species richness, and abundances, but

they do not necessarily conserve patterns of spatial

contagion of species.

Negative values of NRI indicate higher MPD than

expected, and are indicative of phylogenetic evenness,

whereas positive values of NRI indicate phylogenetic

clustering (Webb 2000). To test for significant deviations

of NRI from a null expectation, we calculated a P value

by dividing the number of null communities with MPD

values that were lower than or greater than the observed

by the number of runs þ1. For each plot and each

census, we used this metric to evaluate phylogenetic

structure for tree and seedling assemblages.

Relative contribution of recruited and dead individuals

To assess the relative contribution of recruitment and

mortality to the dynamics of phylogenetic structure over

succession, we calculated NRI for the subset of

survivors, recruited, and dead trees and seedlings at

each census. The reference species pool was the same

regional pool used in the previous analysis. Although

the most logical species pool for measuring phylogenetic

relatedness among dead trees in a particular plot is

arguably the pool of species occurring in that plot, we

used the same species pool across all analyses to obtain

comparable NRI values across all analyses. Based on the

predicted scenario described in the Introduction, we

expected recruited individuals to be more phylogeneti-

cally even than the pool of survivors and dead

individuals to show the opposite pattern. We tested for

the significance of this pattern using a one-tailed, paired

t test for each life stage.

To have a deeper understanding of how tree

recruitment and mortality, as well as future seedling-

to-tree transitions, potentially alter the phylogenetic

composition of the community, we assessed phylogenet-

ic similarity among recruited, dead, and surviving trees

and seedlings within plots, over time. To do so, we

performed an ordination analysis based on the MPD

between all pairs of possible combinations of recruited,

dead, and surviving trees and species across censuses.

We standardized the observed matrix of MPD scores

between pairwise samples by generating 999 random

phylogenetic distance matrices using a randomization

equivalent to the Phylogeny Tip Shuffle null model. For

each null model iteration, we calculated the MPD of all

possible pairs of taxa in one sample to the taxa in the

other to generate a null distribution. This null distribu-

tion was used to calculate a pairwise distance matrix

based on a standardized MPD:

MPDstd ¼ �MPD�MPDrnd

sdMPDrnd

:

We then performed a nonmetric multidimensional

scaling analysis (NMDS) based on these values to

illustrate phylogenetic similarity among all possible

combinations. When compared to an ordination anal-

ysis based on species composition, this analysis reflects

the shared evolutionary history among sites, rather than

the shared presence or absence of species, therefore

emphasizing the similarity in deep phylogenetic struc-

ture, even when no species are shared.

RESULTS


As expected, species classified as second-growth

specialists were significantly clustered within the 203-

species phylogeny (MPDsec ¼ 227.92) when compared

with a null model distribution based on species names

shuffling (95% confidence interval ¼ 230.5–258.6).

Species classified as old-growth specialists showed

tendency to evenness within the same phylogeny

(MPDold ¼ 242.97) when compared with the same type

of null model (95% confidence interval¼225.47–244.90).

Finally, species distribution of generalists within this

phylogeny was within the null expectation (MPDgen ¼229.22, 95% confidence interval ¼ 224.17–254.37).

Temporal changes in phylogenetic structure

Based on species incidence data, we did not observe

any temporal trend in the phylogenetic structure of the

tree assemblages (Fig. 1b). Only Cuatro Rios (CR), the

oldest secondary-forest plot, showed significant evenness

in some censuses. The other plots did not show any

significant pattern. In contrast, abundance-based anal-

yses showed that the phylogenetic evenness of trees


increased over time in all the secondary sites (Fig. 1a). In

the youngest plot, Lindero Sur (LSUR), the abundance-

based NRI indicated that the tree assemblage was

initially phylogenetically clustered, although nonsignif-

icantly, and then became significantly even over time.

The two oldest plots (Lindero El Peje [LEPS] and CR)

were phylogenetically even throughout the whole

monitoring period and showed a decreasing pattern in

NRI until reaching the phylogenetic evenness level of

old-growth forest plots (Fig. 1a). Tirimbina (TIR) was

the only plot with significant phylogenetic clustering of

trees throughout the census period.

Patterns of increasing phylogenetic evenness among

tree individuals over time were due to both recruitment

and mortality processes (Fig. 2). Although the NRI of

recruited and dead trees showed large variation between

consecutive censuses, dead trees were significantly more

closely related than survivors, which, in turn, were

significantly more closely related than recruited trees in

all plots (Fig. 2, Table 2).

Seedling assemblages tended to show greater phylo-

genetic evenness than tree assemblages throughout the

whole monitoring period (Fig. 1c, d). They showed a

sharp decrease in NRI between the third and the fourth

censuses, a phenomenon occurring across all plots, and

particularly pronounced in incidence-based data (Fig.

1d), suggesting that this pattern was the result of species

level population dynamics. Indeed, a previous study

reported the simultaneous colonization of several

canopy palm species that were absent in the previous

censuses in these same plots (Capers et al. 2005).

Because palms (Arecaceae) belong to an early branching

FIG. 1. Temporal changes in the net relatedness index (NRI) based on both species incidence and species abundance of trees infour secondary and two mature forest plots and of seedlings in the four secondary plots in northeastern Costa Rica. Open symbolscorrespond to secondary forest plots, and solid symbols correspond to mature forest plots. Site abbreviations are: LSUR, LinderoSur; TIR, Tirimbina; LEPS, Lindero El Peje; CR, Cuatro Rios; LEPP, LEP primary; and SV, Selva Verde.

* P , 0.05.


angiosperm clade, the arrival of these species into the

community has a large effect on the calculation of

phylogenetic structure. If palms are taken out of the

analysis, the phylogenetic structure of seedling assem-

blages within each plot was stable over time (data not

shown).

In terms of seedling dynamics, the NRI of recruited

and dead individuals showed large variation between

consecutive censuses. Unlike trees, both dead and

recruited individuals tended to be significantly more

phylogenetically clustered than survivors (Fig. 3, Table

2). Recruited seedlings showed the same decrease in NRI

scores as the one observed for the standing seedling

assemblage in all plots, but this was only observed in

LSUR and LEPS for dead individuals (Fig. 3).

In all plots, trees were more phylogenetically dissim-

ilar to seedlings than recruited and dead individuals were

to survivors within each life stage (Fig. 4). For trees,

none of the different groups (deads, recruits, survivors)

overlapped with each other in any of the plots except for

CR (the oldest site), where recruited individuals

overlapped with survivors in the last six censuses (Fig.

4d). LSUR (the youngest site) showed high phylogenetic

similarity among cohorts of dead trees. For seedlings,

dead and recruited individuals exhibited some overlap in

all plots (Fig. 4).

DISCUSSION


The degree of relatedness among early-successional

species has never been addressed quantitatively, partly

because species show a continuous distribution in their

trait values rather than a multimodal shape, and it is

often difficult to place species into a pioneer category

with high certitude (Hubbell and Foster 1992, Wright et

al. 2003). Our study is the first to provide evidence that a

subset of species rigorously classified as second-growth

specialists in northeastern Costa Rica are more closely

related than expected by chance, and that those

classified as old-growth specialists tend to converge in

FIG. 2. Temporal changes in the net relatedness index (NRI) of surviving (solid circles), recruited (open circles), and dead(crossed circles) trees in each of the four study plots, based on species abundance. Note that the points of the graph are shifted oneyear to the right with respect to Fig. 1, since these data refer to recruited, dead, and surviving individuals. See Fig. 1 for siteabbreviations.


their shade-tolerant strategy. Early-successional species

require extreme traits conferring them high dispersal

abilities and high growth rates in disturbed environ-

ments. Seed size (Moles et al. 2005) and wood density

(Chave et al. 2006, Swenson and Enquist 2007), two

traits especially involved in the success of this ecological

syndrome, have consistently shown phylogenetic con-

servatism across a wide range of floras. Since many

other traits have shown similar patterns (Prinzing et al.

2001, Chazdon et al. 2003, Swenson et al. 2007, Kraft et

al. 2010), phylogenetic constraints might have favored

niche conservatism in the pioneer strategy.

Our findings indicate the presence of a phylogenetic

signal in early-successional species, a pattern that,

although often interpreted as niche conservatism, does

not necessarily imply an adaptive process (Losos 2008,

Revell et al. 2008). Phylogenetic niche conservatism

explicitly implies that ‘‘some factor is causing closely

related species to be more ecologically similar that would

be expected by simple Brownian motion descent with

modification’’ (Losos 2008). Our current analysis does

not allow us to draw such a conclusion, but it is a first

step toward a better understanding of the evolution of

the pioneer strategy. Has this ecological syndrome

evolved many times from different lineages, or is it a

conserved strategy that arose from few lineages? The

pioneer strategy probably arose from mature forest

species in response to transient disturbances (Graham

and Janzen 1969, Gomez-Pompa 1971). If our results

can be generalized to larger geographical and taxonomic

scales, early-successional species might have appeared in

few lineages, which diversified thereafter to become

clades showing high niche conservatism, as is the case of

Cecropia. It is important to note that most plant families

are not exclusively composed of early-successional

species, even if they have a predominance of this

syndrome (e.g., Melastomataceae, Urticaceae, or Piper-

aceae; P. Stevenson, personal communication). For

instance, the Urticaceae is known to be a pioneer family

because it contains the genera Cecropia and Pourouma.

Yet Pourouma shows important variation in shade

tolerance, with species such as P. minor commonly

occurring in old-growth forests.

Another issue is geographical scale. Here, we are

focusing on a single lowland tropical forest region, and

it would be ideal to extend our analysis to a larger data

set including the global pool of species occurring in

lowland tropical forests across Central and South

America. A global analysis including all the Neotropical

angiosperms could, however, lead to equivocal interpre-

tations, as it would include species having very different

evolutionary histories (e.g., the Andes orogeny vs. the

Amazon basin formation), and occurring in a variety of

habitats. Investigating further patterns of niche conser-

vatism in species’ successional status will provide key

elements for understanding the evolution of plant

functional traits in relation to community assembly

during succession.

Temporal changes in the phylogenetic structure

of tree assemblages

All plots exhibited increasing phylogenetic evenness

among tree individuals over time, indicating that, at

small spatial scales (1 ha), the probability of co-

occurrence among closely related individuals declines

as succession unfolds. In contrast, the probability of tree

species co-occurrence was unrelated to their phylogenet-

ic background, and did not show any temporal trend.

Such discrepancy between abundance- and incidence-

based analyses can be explained by differences between

individual and species turnover throughout succession.

The first decades of succession typically show high

species turnover due to the replacement of short-lived

pioneers by long-lived pioneers and some shade-tolerant

species (Finegan 1996, Anderson 2007). Then, species

turnover decreases as the pool of potential new

colonizers diminishes, and because late-successional

species have higher longevity than early colonizers. In

contrast, rates of recruitment and mortality stay

elevated, maintaining high individual turnover (Chaz-

don 2003). Our results therefore suggest that rates of

change in species composition in these successional

stands at relatively short timescales are the result of

shifts in species abundance during succession, rather

than of the gain or loss of species (van Breugel et al.

2007, Chazdon et al. 2007). Our monitoring period of 12

years, although quite long for a successional study, may

not be sufficient to discern noticeable patterns of species

turnover that lead to changes in phylogenetic structure.

Letcher (2010) did find an increased phylogenetic

evenness based on species occurrence data across a

TABLE 2. Results of the paired t test analysis comparing netrelatedness index (NRI) values between dead and survivingtrees and seedlings, and recruited and surviving trees andseedlings.

Site and life stage

Dead vs. survivors Recruited vs. survivors

t df t df

LSUR

Trees �3.42** 10 8.82*** 10Seedlings �4.96*** 7 �1.5 7

TIR

Trees �6.14*** 10 8.25*** 10Seedlings 2.00 7 �0.35 7

LEPS

Trees �1.98* 10 6.49*** 10Seedlings �4.03** 7 �1.37 7

CR

Trees �1.79* 10 3.82*** 10Seedlings �6.80*** 7 �2.96 7

Notes: The tests for dead trees and seedlings determinedphylogenetic clustering compared with survivors, whereasrecruited trees and seedlings were tested for phlogeneticevenness compared with survivors. See Table 1 for siteabbreviations.

* P � 0.05; ** P � 0.01; *** P � 0.001.


chronosequence spanning 40 years. Such a long timeperiod covered more successional phases, resulting in a

better ability to detect species turnover. Combining

long-term and demographic data in phylogenetic studiesis therefore a critical step for developing a clear

understanding of the processes driving community

assembly (see also Cadotte et al. 2012, Cavender-Baresand Reich 2012).

In all four secondary plots, the pool of dead trees

consistently showed higher phylogenetic clustering than

the pool of survivors, which, in turn, showed higherphylogenetic clustering than the pool of recruited trees.

These results clearly indicate that both recruitment and

mortality jointly promoted the increased phylogeneticevenness among co-occurring trees observed over

succession. Since we provide evidence that early-

successional species are significantly clustered, and

late-successional species tend to evenness within theangiosperm phylogeny, we can say confidently that the

increased phylogenetic evenness observed here and in

Letcher’s (2010) study is the result of the replacement of

individuals of early-successional species, comprising

many closely related species, by late-successional indi-

viduals belonging to a wider diversity of lineages. This is

well illustrated by the patterns observed in LSUR, where

the repeated death of even-aged cohorts of early-

successional species translated to a high phylogenetic

similarity among all cohorts of dead trees. Cohorts of

recruited trees, in contrast, were more phylogenetically

even in this plot. This pattern was apparent only in this,

the youngest plot, as the other plots represent later

successional stages where tree cohorts are no longer

even-aged. In CR, the oldest plot, recruited trees

overlapped with surviving trees at the last censuses,

indicating that, later in succession, phylogenetic com-

position of recruited trees is not distinguishable from

that of surviving trees. This pattern corroborates

previous findings showing a rapid floristic recovery in

secondary stands in this area (Letcher and Chazdon

2009, Norden et al. 2009).

FIG. 3. Temporal changes in the net relatedness index (NRI) of surviving (solid circles), recruited (open circles), and dead(crossed circles) seedlings in each of the four study plots, based on species abundance. Note that the points of the graph are shiftedone year to the right with respect to Fig. 1, since these data refer to recruited, dead, and surviving individuals. See Fig. 1 for siteabbreviations.


Projecting changes into the future: phylogenetic structure

across life stages

We extended the temporal scale of this study byassessing the community phylogenetic structure at early

life stages too. In the same study area, Letcher (2010)

showed that small stem size classes (0.5–5 cm dbh) weremore phylogenetically even than large stem size classes

(.5 cm dbh). Here, we went further by comparing

seedling and tree assemblages. These two life stagesshowed a clear phylogenetic differentiation, indicating

that the initial colonizers that form early tree cohorts

come from different lineages than the propagules

colonizing a few decades later. Because seedling andsapling layers are comprised of many shade-tolerant

species that do not occur in the canopy of secondary

forests (van Breugel et al. 2007, Norden et al. 2009), thespecies pool of these young life-history stages is likely to

be much larger than the one of adult stages, therefore

exhibiting higher phylogenetic diversity.

Increased phylogenetic evenness in seedlings com-pared to trees indicates that species colonization is an

important ecological process shaping the phylogenetic

structure of successional assemblages. Since species

colonization is directly related to distance from seed

sources, to dispersal limitation, and to the regional

species pool, our results indicate that the surrounding

vegetation matrix has important effects in the temporal

dynamics of local phylogenetic structure. Such land-

scape effects were also reflected in the variation of the

temporal patterns in phylogenetic structure observed

among plots. LSUR and LEPS, the secondary plots

located within the more forested landscape of La Selva

Biological Station, showed the most substantial increase

in phylogenetic evenness during the monitoring period

compared to CR and TIR. As TIR is a young stand (15

years old since abandonment at the beginning of the

monitoring), one would have expected this plot to show

high individual and species turnover (Anderson 2007).

Indeed, species gain exceeded rates of species loss in

LSUR and TIR, but not in LEPS and CR (Chazdon et

al. 2007). Yet, the TIR plot exhibits the highest

phylogenetic clustering among all plots, and very low

changes in phylogenetic evenness over time. This might

be due to the fact that TIR has the longest history of

disturbance and isolation. These observations highlight

FIG. 4. Nonmetric multidimensional scaling (NMDS) analysis illustrating pairwise phylogenetic distances among censuses andplots for surviving (solid symbols), recruited (open symbols), and dead (crossed symbols) trees (circles) and seedlings (squares) foreach plot, based on abundance data. See Fig. 1 for site abbreviations.


the importance of the presence of mature forests within

agricultural matrices.

Temporal variation in seed production also appeared

to promote phylogenetic evenness at the seedling stage.

For instance, the sharp decline in the NRI scores

observed in all seedling assemblages occurred in a short

time frame corresponding to the simultaneous arrival of

palm propagules at the four secondary sites (Capers et

al. 2005). Interannual fluctuations in seed production

typically result in temporal pulses in recruitment,

causing important changes in species richness (Norden

et al. 2007), and thus in the phylogenetic structure of the

seedling community. Moreover, even at late stages of

succession, tree mortality can open new opportunities

for the successful establishment of early-successional

species (van Breugel et al. 2007). These processes

underlie a certain level of unpredictability in the

colonization process, resulting in a complex pattern.

For instance, dead, recruited, and surviving seedlings

did not show any phylogenetic differentiation. It is

important to note that our study focused on recruitment

of individuals into the �20 cm height class. Since early

stages are extremely vulnerable (Harms et al. 2000,

Muller-Landau et al. 2002), nonrandom patterns of

mortality in early establishment could have also

contributed to the observed changes in phylogenetic

structure. Indeed, a recent study shows that phyloge-

netic relatedness decreases among seedlings after the first

year of survival due to negative density dependence

(Metz et al. 2010). Together, these processes lead to

greater temporal fluctuations in phylogenetic structure

in seedling than in tree assemblages.

CONCLUSIONS

To our knowledge, this is the first study to determine

how demographic processes of mortality and recruit-

ment drive changes in the phylogenetic structure of plant

communities. Our results are supported by other

successional studies in ruderal (Knapp et al. 2008) and

zooplankton communities (Helmus et al. 2010), where

species adapted to disturbance were more closely related

than species occurring in non-perturbed ecosystems. We

further show that the phylogenetic structure of second-

ary stands varies at two temporal scales. At short

temporal scales, decreasing relatedness among tree

individuals over succession is the joint result of the

death of closely related early-successional individuals,

and of the recruitment of distantly related late-succes-

sional individuals. At longer temporal scales, the

colonization of propagules belonging to a wide array

of lineages increases phylogenetic evenness in seedlings

compared to trees at any one moment in succession. As

these small stems grow, future tree assemblages are

expected to exhibit higher species richness and phyloge-

netic evenness.

Temporal fluctuation in seed production and species

colonization are key elements contributing to increased

phylogenetic evenness in these reassembling communi-

ties. Under the niche conservatism scenario, studies of

community phylogenetics typically explain phylogenetic

evenness as a consequence of biotic filtering mediated by

competition or by shared pests among closely related

species, and phylogenetic clustering as the result of

environmental filtering (Webb et al. 2002, Kraft et al.

2010). These studies mostly rely on static data from

PLATE 1. A tree-fall gap in a lowland tropical secondary forest at La Selva Biological Station, Costa Rica. Photo credit: R. L.Chazdon.


mature tree communities where some assembly processes

may not be as important as in more dynamic systems.

Recent studies have shown that many processes can

affect phylogenetic structure (Cavender-Bares et al.

2009). For instance, facilitation among distantly related

species appeared to be an important driver of phyloge-

netic diversity in arid plant communities in Central

Mexico (Valiente-Banuet and Verdu 2007) and in

Mediterranean successional communities (Verdu et al.

2009). Here, we further show that other ecological

processes, occurring during the regeneration stage, also

drive successional changes in phylogenetic structure.

Abundance data showed a much stronger signal than

incidence data, suggesting that shifts in species abun-

dance due to recruitment and mortality may show a

concerted pattern of phylogenetic structure long before

the loss and gain of species shows a signal. This finding

is of critical importance, as many community phyloge-

netics studies have drawn conclusions about community

assembly processes based solely upon species occurrence

data. The field of community phylogenetics is a

promising avenue for gaining insight into community

reassembly. Future studies integrating well-resolved

phylogenies (Kress et al. 2009, Gonzalez et al. 2010),

demographic data and long-term monitoring will

provide increased sensitivity for unraveling the mecha-

nisms that determine community assembly in succes-

sional forests.

ACKNOWLEDGMENTS

We thank Sebastian Gonzalez for insightful discussionsabout community phylogenetics. David Ackerly, Daniel Cade-na, and two anonymous reviewers provided helpful commentsthat improved earlier versions of the manuscript. This researchwas supported by grants from the Andrew W. MellonFoundation, NSF DEB-0424767, NSF DEB-0639393, and theUniversity of Connecticut Research Foundation.

LITERATURE CITED

Anderson, K. J. 2007. Temporal patterns in rates of communitychange during succession. American Naturalist 169:780–793.

Blomberg, S. P., T. Garland, and A. R. Ives. 2003. Testing forphylogenetic signal in comparative data: behavioral traits aremore labile. Evolution 57:717–745.

Cadotte, M. W., R. Dinnage, and D. Tilman. 2012. Phyloge-netic diversity promotes ecosystem stability. Ecology93(Supplement):S223–S233.

Capers, R. S., R. L. Chazdon, A. R. Brenes, and B. V.Alvarado. 2005. Successional dynamics of woody seedlingcommunities in wet tropical secondary forests. Journal ofEcology 93:1071–1084.

Cavender-Bares, J., K. H. Kozak, P. V. A. Fine, and S. W.Kembel. 2009. The merging of community ecology andphylogenetic biology. Ecology Letters 12:693–715.

Cavender-Barres, J., and P. B. Reich. 2012. Shocks to thesystem: community assembly of the oak savanna in a 40-yearfire frequency experiment. Ecology 93(Supplement):S52–S69.

Chave, J., H. C. Muller-Landau, T. R. Baker, T. A. Easdale, H.Ter Steege, and C. O. Webb. 2006. Regional and phyloge-netic variation of wood density across 2456 Neotropical treespecies. Ecological Applications 16:2356–2367.

Chazdon, R. L. 2003. Tropical forest recovery: legacies ofhuman impact and natural disturbances. Perspectives inPlant Ecology, Evolution and Systematics 6:51–71.

Chazdon, R. L. 2008. Chance and determinism in tropicalforest succession. Pages 384–408 in W. P. Carson and S. A.Schnitzer, editors. Tropical forest community ecology.Blackwell, Malden, Massachusetts, USA.

Chazdon, R. L., S. Careaga, C. O. Webb, and O. Vargas. 2003.Community and phylogenetic structure or reproductive traitsof woody species in wet tropical forests. Ecology 73:331–348.

Chazdon, R. L., A. Chao, R. K. Colwell, S.-Y. Lin, N. Norden,S. G. Letcher, D. B. Clark, B. Finegan, and J.-P. Arroyo.2011. A novel statistical method for classifying habitatgeneralists and specialists. Ecology 92:1332–1343.

Chazdon, R. L., S. G. Letcher, M. van Breugel, M. Martinez-Ramos, F. Bongers, and B. Finegan. 2007. Rates of change intree communities of secondary Neotropical forests followingmajor disturbances. Philosophical Transactions of the RoyalSociety of London B 362:273–289.

Chazdon, R. L., C. A. Peres, D. Dent, D. Sheil, A. E. Lugo, D.Lamb, N. E. Stork, and S. E. Miller. 2009. The potential forspecies conservation in tropical secondary forests. Conserva-tion Biology 6:1406–417.

FAO [Food and Agriculture Organization]. 2005. The state ofthe world’s forests. FAO, Rome, Italy.

Finegan, B. 1996. Pattern and process in neotropical secondaryrain forests: the first 100 years of succession. Trends inEcology and Evolution 11:119–124.

Fritz, S. A., and A. Purvis. 2010. Selectivity in mammalianextinction risk and threat types: a new measure ofphylogenetic signal strength in binary traits. ConservationBiology 24:1042–1051.

Gomez-Pompa, A. 1971. Posible papel de la vegetacionsecundaria en la evolucion de la flora tropical. Biotropica3:125–135.

Gonzalez, M. A., A. Roger, E. A. Courtois, F. Jabot, N.Norden, C. E. T. Paine, C. Baraloto, C. Thebaud, and J.Chave. 2010. Shifts in species and phylogenetic diversitybetween sapling and tree communities indicate negativedensity dependence in a lowland rain forest. Journal ofEcology 98:137–146.

Graham, A., and D. M. Janzen. 1969. Studies in Neotropicalpaleobotany: I. The Oligocene communities of Puerto Rico.Annals of the Missouri Botanical Garden 56:308–357.

Hardy, O. J. 2008. Testing the spatial phylogenetic structure oflocal communities: statistical performances of different nullmodels and test statistics on a locally neutral community.Journal of Ecology 96:914–926.

Harms, K. E., R. Condit, S. P. Hubbell, and R. B. Foster. 2001Habitat associations of trees and shrubs in a 50-haneotropical forest plot. Journal of Ecology 89:947–959.

Helmus, M. R., W. Keller, M. J. Paterson, N. D. Yan, C. H.Cannon, and J. A. Rusak. 2010. Communities contain closelyrelated species during ecosystem perturbance. EcologyLetters 13:162–174.

Hubbell, S. P. 2001. The unified neutral theory of biodiversityand biogeography. Princeton University Press, Princeton,New Jersey, USA.

Hubbell S. P. 2005. Neutral theory in community ecology andthe hypothesis of functional equivalence. Functional Ecology19:166–172.

Hubbell, S. P., and R. B. Foster. 1992. Short-term dynamics ofa neotropical forest: why ecological research matters totropical conservation and management. Oikos 63:48–61.

ITTO [International Tropical Timber Organization]. 2002.ITTO guidelines for the restoration, management andrehabilitation of degraded and secondary forests. ITTOPolicy Development Series Number 13. ITTO, Yokohama,Japan.

Jabot, F., and J. Chave. 2009. Inferring the parameters of theneutral theory of biodiversity using phylogenetic informa-tion, and implications for tropical forests. Ecology Letters12:239–248.

Kembel, S. W., P. D. Cowan, M. R. Helmus, W. K. Cornwell,H. Morlon, D. D. Ackerly, S. P. Blomberg, and C. O. Webb.


2010. Picante: R tools for integrating phylogenies andecology. Bioinformatics 26:1463–1464.

Knapp, S., I. Kuhn, O. Schweiger, and S. Klotz. 2008.Challenging urban species diversity: contrasting phylogeneticpatterns across plant functional groups in Germany. EcologyLetters 11:1054–1064.

Kraft, N. J. B., and D. D. Ackerly. 2010. Functional trait andphylogenetic tests of community assembly across spatialscales in an Amazonian forest. Ecological Monographs80:401–422.

Kress, W. J., D. L. Erickson, F. A. Jones, N. G. Swenson, R.Perez, O. Sanjur, and E. Bermingham. 2009. Plant DNAbarcodes and a community phylogeny of a tropical forestdynamic plot in Panama. Proceedings of the NationalAcademy of Science USA 106:18621–18626.

Letcher, S. G. 2010. Phylogenetic structure of angiospermcommunities during tropical forest succession. Proceedings ofthe Royal Society B 277:97–104.

Letcher, S. G., and R. L. Chazdon. 2009. Rapid recovery ofbiomass, species richness and species composition in a forestchronosequence in northeastern Costa Rica. Biotropica41:608–617.

Losos, J. B. 2008. Phylogenetic niche conservatism, phyloge-netic signal and the relationship between phylogeneticrelatedness and ecological similarity among species. EcologyLetters 11:995–1007.

McDade, L., K. S. Bawa, H. Hespenheide, and G. S.Hartshorn. 1994. La Selva: ecology and natural history of aNeotropical rain forest. University of Chicago Press,Chicago, Illinois, USA.

Metz, M. R., W. P. Sousa, and R. Valencia. 2010. Widespreaddensity-dependent seedling mortality promotes species coex-istence in a highly diverse Amazonian rainforest. Ecology91:3675–3685.

Moles, A. T., D. D. Ackerly, C. O. Webb, J. C. Tweddle, J. B.Dickie, and M. Westoby. 2005. A brief history of seed size.Science 307:576–580.

Muller-Landau, H. C., et al. 2006. Testing metabolic ecologytheory for allometric scaling of tree size, growth andmortality in tropical forests. Ecology Letters 9:575–588.

Muller-Landau, H. C, S. J. Wright, O. Calderon, S. P. Hubbell,and R. B. Foster. 2002. Assessing recruitment limitation:concepts, methods and case-studies from a topical forest.Pages 35–53 in D. J. Levey, W. R. Silva, and M. Galetti,editors. Seed dispersal and frugivory. CAB International,Wallingford, UK.

Norden, N., J. Chave, A. Caubere, P. Chatelet, N. Ferroni,P.-M. Forget, and C. Thebaud. 2007. Is temporal variation ofseedling communities determined by environment or by seedarrival? A test in a neotropical forest. Journal of Ecology95:507–516.

Norden, N., R. L. Chazdon, A. Chao, Y. H. Jiang, and B.Vilchez-Alvarado. 2009. Resilience of tropical rain forests:tree community reassembly in secondary forests. EcologyLetters 12:385–394.

Prinzing, A., W. Durka, S. Klotz, and R. Brandl. 2001. Theniche of higher plants: evidence for phylogenetic conserva-tism. Proceedings of the Royal Society B 268:2383–2389.

Proyecto Flora Digital de La Selva. 2009. Lista de plantasvasculares. Estacion Biologica La Selva. Organizacion deEstudios Tropicales (OET), Sarapiquı, Costa Rica: www.ots.ac.cr/local/florula

R Development Core Team. 2010. R: A language andenvironment for statistical computing. Version 2.11.1. RFoundation for Statistical Computing, Vienna, Austria.http://www.R-project.org

Rees, M., R. Condit, M. Crawley, S. Pacala, and D. Tilman.2001. Long-term studies of vegetation dynamics. Science293:650–654.

Revell, L. M., L. J. Harmon, and D. C. Collar. 2008.Phylogenetic signal, evolutionary process, and rate. System-atic Biology 57:591–601.

Swenson, N. G. 2009. Phylogenetic resolution and quantifyingthe phylogenetic diversity and dispersion of communities.PLoS ONE 4(2):e4390.

Swenson, N. G., and B. J. Enquist. 2007. Ecological andevolutionary determinants of a key plant functional trait:wood density and its community-wide variation acrosslatitude and elevation. American Journal of Botany91:451–459.

Swenson, N. G., B. J. Enquist, J. Thompson, and J. K.Zimmerman. 2007. The influence of spatial and size scale onphylogenetic relatedness in tropical forest communities.Ecology 88:1770–1780.

Swenson, N. G., et al. 2012. Phylogenetic and functional alphaand beta diversity in temperate and tropical tree communi-ties. Ecology 93(Supplement):S112–S125.

Valiente-Banuet, A., and M. Verdu. 2007. Facilitation canincrease the phylogenetic diversity of plant communities.Ecology Letters 10:1029–1036.

Vamosi, S. M., S. B. Heard, C. Vamosi, and C. O. Webb. 2009.Emerging patterns in the comparative analysis of phyloge-netic community structure. Molecular Ecology 18:572–592.

van Breugel, M., F. Bongers, and M. Martınez-Ramos. 2007.Species dynamics during early secondary forest succession:recruitment, mortality and species turnover Biotropica35:610–619.

Verdu, M., and J. G. Pausas. 2007. Fire drives phylogeneticclustering in Mediterranean Basin woody plant communities.Journal of Ecology 95:1316–1323.

Verdu, M., P. J. Rey, J. M. Alcantara, G. Siles, and A.Valiente-Banuet. 2009. Phylogenetic signatures of faciliationand competition in successional communites. Journal ofEcology 97:1171–1180.

Warwick, R. M., and K. R. Clarke. 1998. Taxonomicdistinctness and environmental assessment. Journal ofApplied Ecology 35:532–543.

Webb, C. O. 2000. Exploring the phylogenetic structure ofecological communities: an example for rain forest trees.American Naturalist 156:145–155.

Webb, C. O., D. D. Ackerly, and S. W. Kembel. 2008.Phylocom: software for the analysis of phylogenetic commu-nity structure and trait evolution. Bioinformatics 24:2098–2100.

Webb, C. O., D. D. Ackerly, M. A. McPeek, and M. J.Donoghue. 2002. Phylogenies and community ecology.Annual Review of Ecology and Systematics 33:475–505.

Webb, C. O., and M. J. Donoghue. 2005. Phylomatic: treeassembly for applied phylogenetics. Molecular EcologyNotes 5:181–183.

Westoby, M., D. S. Falster, A. T. Moles, P. A. Vesk, and I. J.Wright. 2002. Plant ecological strategies: some leadingdimensions of variation between species. Annual Review ofEcology and Systematics 33:125–159.

Wikstrom, N., V. Savolainen, and M. W. Chase. 2001.Evolution of the angiosperms; calibrating the family tree.Proceedings of the Royal Society B 268:2211–2220.

Wright, S. J., H. C. Muller-Landau, R. Condit, and S. P.Hubbell. 2003. Gap dependent recruitment, realized vitalrates, and size distributions of tropical trees. Ecology84:3174–3185.


demographic drivers of successional changes in …swensonn/nordenetal_2012_ecology.pdfecology, 93(8)...

Documents