You are on page 1of 12

Journal of Biogeography (J. Biogeogr.

) (2011) 38, 394–405

ORIGINAL Phylogenetic diversity, trait diversity and


ARTICLE
niches: species assembly of ferns along a
tropical elevational gradient
Jürgen Kluge1,2* and Michael Kessler1

1
Systematic Botany, University of Zurich, ABSTRACT
Zollikerstrasse 107, CH-8008 Zurich,
Aim To analyse the structure of pteridophyte assemblages, based on phylogenetic
Switzerland, 2Faculty of Geography, University
of Marburg, Deutschhausstrasse 10, D-35032
relatedness and trait properties, along an elevational gradient. Ecological theory
Marburg, Germany predicts that co-occurring species may be: randomly selected from a regional
pool; ecologically sorted so that they are functionally different hence resulting in
reduced competition (overdispersion); or functionally similar as an adaptation to
specific ecological conditions (clustering).
Location Braulio Carrillo National Park and Cerro de la Muerte, Costa Rica,
Central America.
Methods We used an empirical dataset of the quantitative pattern of species
occurrences and individual numbers of ferns within 156 plots along a tropical
elevational gradient to test whether directed ecological sorting might cause
deviations in patterns of trait and phylogenetic diversity. Mean pairwise distances
of species based on phylogenetic and trait properties were compared with two
different sets of null assemblages, one maintaining species frequency distributions
(constrained) and one not (unconstrained).
Results Applying different null models resulted in varying degrees of
overdispersion and clustering, but overall patterns of deviation from random
expectations remained the same. Contrary to theoretical predictions, phylogenetic
and trait diversity were relatively independent from one another. Phylogenetic
diversity showed no patterns along the elevational gradient, whereas trait diversity
showed significant trends for epiphytes.
Main conclusions Under stressful environmental conditions (drought at low
elevations and frost at high elevations), epiphytic fern assemblages tended to be
clustered with respect to trait characteristics, which suggests environmental
filtering. Conversely, under less extreme environmental conditions (middle of the
transect), the sorting was biased towards high differentiation (overdispersion),
presumably because of interspecific competition and trait shifts among closely
*Correspondence: Jürgen Kluge, Faculty of related species (character displacement).
Geography, University of Marburg,
Keywords
Deutschhausstrasse 10, D-35032 Marburg,
Germany. Adaptation, Costa Rica, environmental filtering, ferns, niche, phylogenetic
E-mail: klugejuergen@gmx.de diversity, species assemblage, trait diversity, tropical forest.

and composition brought about by global change (Fargione


INTRODUCTION
et al., 2007). Biological diversity includes a wide range of
Understanding how the diversity of ecological communities attributes, ranging from genetic diversity within populations to
has evolved and how this diversity influences ecosystem ecosystem diversity on a global scale (Magurran, 1988).
functionality is one of the central topics of ecological research, Because species are the most readily sampled entities, the vast
especially with regard to the changes in assemblage diversity majority of studies on the diversity and structure of species

394 www.blackwellpublishing.com/jbi ª 2010 Blackwell Publishing Ltd


doi:10.1111/j.1365-2699.2010.02433.x
Phylogenetic and trait diversity of ferns

assemblages have used species as focal units (Lomolino et al., number, a pattern called ‘overdispersion’ (Moulton & Pimm,
2006). However, understanding processes that generate vari- 1987) (Fig. 1).
ation in diversity requires criteria to measure diversity that go Second, the environment may set limits on the occurrence
beyond simple species richness. As newly evolved taxa tend to of species unsuited to these ambient conditions. Such
be ecologically similar (Wiens & Graham, 2005), there may be ‘environmental filters’ might constrain species composition
a link between the phylogenetic relatedness of species within an to a restricted set of possible characteristics (Box, 1981;
assemblage, traits of these species, and the ecological processes Keddy, 1992; Weiher & Keddy, 1995; de Bello et al., 2006)
that determine their co-occurrence (Kraft et al., 2007). Along and admit only species that are viable under the given
this line of thought, species assemblages may show non- environmental conditions (e.g. climate, disturbance regimes)
random patterns of species co-occurrences (reviewed by Webb and biotic interactions (e.g. competitive exclusion) (Smith
et al., 2002), and the degree of relatedness between co- et al., 1994). In this way, the diversity of characters is
occurring species in terms of phylogenetic and trait similarity limited by the local niche space (Dı́az & Cabido, 2001;
may shed light on prevailing ecological processes such as Grime, 2002; Lavorel & Garnier, 2002). This process might
environmental filtering (Weiher & Keddy, 1995) and niche thus result in trait diversity levels lower than those of
differentiation (Stubbs & Wilson, 2004). random assemblages, a pattern called ‘clustering’ (Weiher &
As species usually differ from each other with respect to Keddy, 1995) (Fig. 1).
their position in the phylogenetic tree, as well as in their Accordingly, if there is directional ecological sorting (sensu
morphological and thus functional characters, the range of Ackerly, 2003), the direction itself may shed light on the
species properties tends to increase with species richness mechanism of species assembly (Fridley, 2001; Mouquet et al.,
(Janzen, 1985; Sax et al., 2007) (Fig. 1). This relationship is not 2002). In general, both types of sorting are well supported by
linear but asymptotic because, as new species are added to an both theoretical and empirical studies, but it has not yet been
assemblage, the probability of their carrying new characteris- explored how they affect the relationship of species richness
tics declines as the number of species increases. The shape of and phylogenetic/trait diversity along natural diversity gradi-
this asymptotic curve might be modified by two contrasting ents. A priori, one might expect competition to be stronger in
processes (Weiher & Keddy, 1995). species-rich assemblages, and environmental filtering to be
First, the probability of a species persisting may increase if more prevalent in species-poor assemblages under extreme
it is distinct from other species in an assemblage. This will environmental conditions. Alternatively, however, species-rich
result in a more efficient exploitation of the available assemblages may contain a higher proportion of functionally
resources, which in turn will reduce competitive interactions redundant (and hence similar) species, whereas species-poor
(MacArthur, 1958; Diáz & Cabido, 1997; Fridley, 2001; ones may contain distinct species exploiting different niches
Hooper et al., 2005). Under such a regime, competition (Yachi & Loreau, 2007). Thus there are various reasons to
should favour species with different characters and thus from expect that the phylogenetic/trait diversity–species richness
different lineages, resulting in a level of diversity above that relationship might differ from randomness (Weiher & Keddy,
of randomly selected assemblages with the same species 1995).
On the assumption that traits are conserved within lineages,
phylogenetic and trait diversity patterns should vary simulta-
neously. Disagreements of patterns along gradients may give
further insight into assembling processes beyond competition
and filtering (Webb et al., 2002). Species not only may be
selected into assemblages, but also may evolve while persisting
within an assemblage, thus altering phylogenetic and trait
structures in different ways (Cavender-Bares et al., 2009).
In the present study we assess the structure of fern
assemblages with respect to phylogenetic relatedness and trait
characteristics of co-occurring species using the species-
abundance distribution of ferns and lycophytes along a
tropical elevational gradient with a strong species-richness
pattern. Species richness was highest at mid-elevations, creat-
ing a symmetrical hump along the elevational gradient, as also
found at the same study site by Cardelús et al. (2006) and
Watkins et al. (2006). Because the major life forms of ferns,
Figure 1 Theoretical curves showing the hypothetical relation-
ships between local species richness and trait diversity. The con- terrestrials and epiphytes not only differ in their elevational
tinuous line shows random sorting from a regional pool; the upper and morphological diversity patterns (Kessler, 2001; Kluge &
dashed line represents a sorting towards higher niche differentia- Kessler, 2007), but also occupy different realms within the
tion and reduced competition, the lower dashed line a sorting vegetation assemblages, we conducted our analyses separately
towards adaptation and convergence. for each life form.

Journal of Biogeography 38, 394–405 395


ª 2010 Blackwell Publishing Ltd
J. Kluge and M. Kessler

MATERIALS AND METHODS project (http://www.tolweb.org), a regularly updated collabo-


rative effort of taxonomic specialists on the respective groups
Vegetation sampling of organisms.
For the assessment of the phylogenetic relationships among
Our analyses are based on distributional data for 477 fern species species, several authors have suggested taking the length of the
along an elevational gradient (100–3400 m a.s.l.) with strong branches of a taxonomic cladogram (e.g. Vane-Wright et al.,
associated changes in environmental conditions in undisturbed 1991; Faith, 1992, 1995), but unfortunately such fully resolved
tropical rain forest in Costa Rica, Central America. Species cladograms are rarely available (Crozier, 1992; Izsák & Papp,
assemblages were sampled within 156 plots with a regular square 2000; Ricotta, 2005; Schweiger et al., 2008). To compensate for
shape of 20 · 20 m2. Two study sites were combined: Volcan the lack of such phylogenetic information, here we used the
Barva (100–2800 m) and Cerro de la Muerte (2700–3400 m). topological distance of species within the phylogenetic tree of
Although these two gradients are c. 50 km apart, previous the whole species pool by counting the nodes separating each
analysis revealed that there is no abrupt change in species two species within an assemblage (mean pairwise distance:
compositions between these sites (Kluge et al., 2008), so both Izsák & Papp, 2000; Webb, 2000; Ricotta, 2002), leading to the
can be treated in combination. All species were counted distance matrix mentioned above, by using the R package
separately for terrestrials and epiphytes. For terrestrial species, ‘picante’ (Kembel et al., 2009). Note that this measure does not
we avoided special substrates such as dead wood and stones to consider the true length of cladogram branches, therefore this
minimize the possible influence of habitat heterogeneity. The calculation is an oversimplification.
species-richness pattern is strongly hump-shaped with the
highest values at mid-elevations, reaching a maximum of 68
species per 400 m2 plot, and falling to 10 and 15 species per plot Trait data
at the upper and lower ends of the gradient, respectively (for In the same manner, the structure of assemblages can be
further details see Kluge et al., 2006). analysed by replacing the phylogenetic distance matrix by a
species distance matrix based on trait properties of the species.
General analysis procedure Morphological characteristics of the species were compiled
from the literature (Moran & Riba, 1995; Rojas, 1996, 1997,
Species-assembly patterns were assessed by plot-wide phylo- 2001a,b, 2002a,b) complemented by the study of herbarium
genetic and trait relationships between the species present in material. The traits, their scale of measurement, and their
the respective assemblages. Assemblages were defined as the presumed functional importance are listed in Table 1. Some of
species co-occurring in each study plot. This calculation is these characteristics, especially indument density values and
based on distance matrices, where each pair of species present laminar thickness, were rather subjective and variable; to avoid
in all assemblages found within this study (species pool) was a pretence to accuracy, we used an ample class division (see
assigned a distance value (Euclidean) with respect to their further details in Kluge & Kessler, 2007). Because traits are
phylogenetic and trait similarity. usually expressed on different scales, trait data needed to be
From the whole set of pairwise distances, we calculated the standardized to avoid traits with higher numeric values
mean of all pair values for those species that were members of overriding traits with lower values. We standardized the trait
the respective assemblage (mean distance). To account for values to means of 0 and a standard deviation of 1 so that all
abundance distributions within the assemblages and to give traits had the same weight.
lower weight to rare and possibly ‘erratic’ species, the distance
values were weighted according to the abundance of the
respective species. In this way, each assemblage was assigned a Null assemblages
value of its mean phylogenetic distance (MPDemp) and mean The choice of a null model is essential for interpretation of the
trait distance (MTDemp), respectively. As this distance is a comparison of empirical and randomized species assemblages
measure for the phylogenetic/trait diversity of an assemblage, (Gotelli & Graves, 1996; Webb et al., 2002; Kembel & Hubbell,
MPD and MTD may also be read as ‘mean phylogenetic 2006; Swenson, 2009). We contrast the patterns gained from
diversity’ and ‘mean trait diversity’, respectively. two different null models that imply different assumptions
To assess whether such empirical assemblage patterns along about the ability of species to occur at a certain site.
the elevational gradient of both phylogenetic and trait The first null model had fewer restrictions, and rests on the
relationships differed significantly from random assemblages, assumption that each species is able to colonize a given plot
we tested the empirical values against the values of 1000 with equal probability. In the construction of these null
randomly constructed assemblages. assemblages, the species were chosen randomly without
replacement from the whole species pool, so the species
richness of each plot and the total abundance of all species
Phylogenetic data
across all plots were kept constant, but not the abundance of
We constructed the phylogenetic tree by using data on each species in each plot. As species occurrence frequencies in
phylogenetic relationships provided by the Tree of Life web the randomized assemblages were not forced to be equal to the

396 Journal of Biogeography 38, 394–405


ª 2010 Blackwell Publishing Ltd
Table 1 Morphological traits of ferns used in this study according to information from Moran & Riba (1995), Rojas (1996, 1997, 2001a,b, 2002a,b) and herbarium studies, and their assignment
into classes.

Trait Scale of measurement Description Class Source* Functional importance


Rhizome type Ordinal Absent/unknown 0 a Competition for space/light, especially in

ª 2010 Blackwell Publishing Ltd


Upright to short-creeping 1 a dense communities (Kluge & Kessler, 2007)
Creeping 2 a

Journal of Biogeography 38, 394–405


(Stout) tree-like 3 a
Laminar dissection Ordinal Entire 0 a Adaptations to harsh environments,
Once-pinnate (to once-pinnate-pinnatifid) 1 a especially marked temperature and
Twice- or more-pinnate 2 a humidity oscillations (Kramer et al., 1995;
Halloy & Mark, 1996)
Laminar morphology Ordinal Monomorphic 1 a Dimorphism facilitates spore dispersal
Slightly dimorphic 2 a (Kramer et al., 1995)
Dimorphic (including heterophyllous) 3 a
Laminar thickness Ordinal Thin (membranaceous to thin herbaceous) 1 a Thick leaves against harsh environments e.g.
Medium (herbaceous to paper-like) 2 a frost, drought (Dubuisson et al., 2003)
Thick (thick paper-like to very coriaceous, 3 a
stiff)
Scale and hair density Ordinal Absent (0%) 0 b Dense indument as protection against
Very sparse/weak (> 0–1%) 1 b desiccation, solar radiation, herbivory or
Sparse (1–10%) 2 b gas exchange blocking water films
Numerous (10–50%) 3 b (Ehleringer & Mooney, 1978; Smith &
Very numerous to covering surface (> 50%) 4 b McClean, 1989; Halloy & Mark, 1996)
Hydathodes Binary Absent 0 a Hydathodes facilitate effective water
Present 1 a transport and secretion of surplus
minerals (Kramer et al., 1995)
Indusium Binary Absent 0 a Indusia as mechanical defence against high
Present 1 a amounts of falling and running water,
protection of spores (Kramer et al., 1995)
Buds Binary Absent 0 a Buds as vegetative reproduction mode,
Present 1 a especially in wet environments (Mickel,
1976; Koptur & Lee, 1993)

*Source of information: a, literature information verified by personal observations; b, personal observations.


Phylogenetic and trait diversity of ferns

397
J. Kluge and M. Kessler

empirical frequencies, we called this null model the ‘uncon- the co-occurring species are more closely related than expected
strained null model’. by chance (clustering), whereas positive values and high
The second model kept constant not only the species quantiles (P > 0.95) indicate species assemblages where the
richness of each plot, but also the abundance of each species co-occurring species are less closely related than expected by
across all plots. This model assumes that the ability of chance (overdispersion). Values close to zero thus indicate
species to colonize a plot is a direct function of its frequency in random assemblages.
the whole dataset (Kembel & Hubbell, 2006). This was
achieved using the independent swap algorithm (Gotelli,
Pseudoreplication and spatial autocorrelation
2000; Gotelli & Entsminger, 2001), also called the ‘checker-
board swap’. This algorithm searches for submatrices within Each plot is assigned a significance value stating whether its
the original species · plot matrix of the form (0,1)(1,0) (where phylogenetic structure differs from that of the 1000 random
0 and 1 represent absent and present species, respectively; note assemblages. However, when considering a set of such
that the respective cells of the submatrix need not be adjacent statistical inferences simultaneously, hypothesis tests that
in the original matrix), and swaps the entries of these cells so incorrectly reject the null hypothesis are more likely to occur
that species richness of each plot and total frequency of the (that is, a form of pseudoreplication occurs; Benjamini &
species across all plots are maintained. The number of such Yekutieli, 2001). To cope with this, a stronger level of evidence
swappings for each generated null assemblage (30,000) exceeds is required for an individual test to be stated as significant, to
the number of filled cells within the original matrices (c. 2000 compensate for the number of tests being made. We applied
for terrestrial species, c. 4200 for epiphytic species), in order to the false discovery rate (FDR) analysis to correct for such
ensure that the resulting randomized matrix is sufficiently pseudoreplicated comparisons by computation of a P value
different from the empirical one (Lehsten & Harmand, 2006). cut-off (Storey, 2002).
As this model contrasts with the assumptions of the former Additionally, spatial autocorrelation may affect the percep-
model concerning species frequency distributions, hereafter we tion of species distribution data along spatial or other
call it the ‘constrained null model’. gradients, because these distributions are inherently spatially
structured, thus adjacent locations tend to be more similar in
terms of ecological properties and processes than expected by
Structure of assemblages
chance (Legendre, 1993). We used simultaneous autoregressive
We calculated the mean pairwise phylogenetic and trait (SAR; Bivand et al., 2008) models to expand traditional linear
distance of all species pairs occurring in an assemblage based models (e.g. based on ordinary least squares, OLS) with an
on the empirical data (MPDemp, MTDemp). To determine if additional spatial term (spatial weight matrix), which incor-
the composition differed from randomness, we compared the porates the spatial autocorrelation structure of the data.
observed empirical assemblage values with values of ran- Neighbourhood weights for the construction of the spatial
domly generated null assemblages (Gotelli & Graves, 1996). weight matrix need to be defined, with the aim of reducing the
For 1000 such null assemblages, we calculated the mean Moran’s I of the OLS models to almost zero and not significant
and standard deviation (MPDnull, sdMPDnull, MTDnull, in the SAR models (Dormann et al., 2007). The resulting SAR
sdMTDnull). models are able to reduce the spatial pattern of model residuals
The phylogenetic and trait diversity measured with this with independently distributed errors in regression models
method is, to a certain degree, correlated with species richness (Kissling & Carl, 2008).
(for assessment of the performance of several phylogenetic All calculations were performed using R (R Development
diversity indices, see Schweiger et al., 2008). We were inter- Core Team, 2009), with package picante (Kembel et al., 2009)
ested to determine whether an empirical assemblage value for the analysis of phylogenetic and trait structures; package
differed from random expectation given the species richness of qvalue (Dabney et al., 2009) for performing the FDR; and
an assemblage. Thus we calculated the standardized effect size package spdep (Bivand, 2006) to run the simultaneous
(Gotelli & Rohde, 2002) of the empirical phylogenetic autoregressive models.
distances relative to the phylogenetic distances of the null
assemblages (MPDstand and MTDstand, respectively), which is
RESULTS
defined as
 Applying SAR models to correct linear models for spatial
Metricstand ¼ Metricemp  Metricnull =sd Metricnull
autocorrelation eliminated the error terms in Moran’s I
This measure is equivalent to )1 times the net relatedness present in the OLS analyses (see Appendix S1 in Supporting
index (NRI, Kembel et al., 2009), which has been proposed as Information). Hereafter, all reports on explained variances and
a measure for tree-wide phylogenetic clustering and overdi- the respective significances result from the SAR models only.
spersion of species within an assemblage (Webb, 2000; Kembel Patterns of phylogenetic and trait diversity (MPD, MTD)
& Hubbell, 2006), and is independent of species richness were significantly correlated, more strongly so for terrestrial
(Swenson, 2009). Negative MPDstand and MTDstand values and species (R = 0.66, P < 0.001) than for epiphytic species
low quantiles (P < 0.05) indicate species assemblages where (R = 0.34, P < 0.001).

398 Journal of Biogeography 38, 394–405


ª 2010 Blackwell Publishing Ltd
Phylogenetic and trait diversity of ferns

Figure 2 Mean phylogenetic diversity (MPD) and mean trait diversity (MTD) of fern assemblages along an elevational gradient in Costa
Rica, sorted by plot species richness. Each dot marks an individual plot (terrestrials, n = 156; epiphytes, n = 155). Regression lines with
logarithmic term and R2 are given; **P < 0.01; ***P < 0.001.

The arrangement of MPD and MTD values by species paring the empirical with the simulated values of random
richness (Fig. 2) showed that these relationships were not species distributions showed that phylogenetic and trait
linear, but asymptotically reached a plateau at high levels of structure were highly variable among plots (Fig. 3). Most
species richness. All relationships were significant (P < 0.01), values fell within the 90% confidence intervals of the null
and R2 values of the regression models ranged from 0.30 to models, but the standardized values indicate clustering (neg-
0.54, emphasizing the asymptotic fit for the model curves. R2 ative) or overdispersion (positive) of assemblages with respect
values were higher for epiphytic than for terrestrial species. to phylogenetic and trait variation. As expected, the uncon-
Standardized values of phylogenetic and trait diversity strained null model had higher proportions of plots tending
(MPDstand, MTDstand) were largely independent from variation towards clustering or overdispersion than the constrained null
in species richness (Table 2) for both randomization proce- model (Fig. 3).
dures, with the exception of values based on MTD of When the elevational trends of MPDstand and MTDstand were
epiphytes, in which relationships were significant but rather analysed for all plots together, the significance level had to be
weak (R2 < 0.15). Therefore we relied on these standardized corrected due to multiple hypothesis testing (pseudoreplica-
values as an independent measure of species assembly. tion). While a considerable number of plots showed significant
Mean standardized values obtained using the unconstrained deviations from null expectations (dark grey columns in
null model differed more strongly from zero than those Fig. 3), applying FDR left no significant deviations for any
obtained using the constrained null model (Table 2). Com- assemblage. Along the elevational gradient, clear and signifi-
cant trends were found only for epiphytes with respect to
MTDstand (Fig. 4). Trends in MTDstand of epiphytes were
Table 2 Results of regression analyses of standardized values of
mean phylogenetic diversity (MPDstand) and mean trait diversity
hump-shaped in both randomization procedures, indicating
(MTDstand) and fern species richness in 156 forest plots in Costa that epiphytic assemblages had higher trait diversity than
Rica, separated for terrestrial and epiphytic species. random assemblages at the middle of the gradient (1000–
2000 m), and correspondingly lower trait diversity at the ends
Randomization Diversity Species Mean towards lowlands and the mountain tops.
procedure measure group Slope R2 P value

Unconstrained MPD Terrestrial 0.008 0.00 0.591 )0.18 DISCUSSION


Epiphytic 0.006 0.01 0.261 )0.56
MTD Terrestrial 0.033 0.02 0.110 0.14 Our study compares the phylogenetic and trait structure of
Epiphytic 0.060 0.15 < 0.001 1.05 ferns within the same assemblages to distinguish between
Constrained MPD Terrestrial 0.008 0.00 0.630 )0.06 ecological sorting of species based on their traits or their
Epiphytic 0.012 0.01 0.144 )0.09 evolutionary legacies. Because phylogenetic groups are gener-
MTD Terrestrial 0.009 0.00 0.638 )0.03 ally characterized by suites of traits, a close relationship of MPD
Epiphytic 0.027 0.07 0.001 0.11 and MTD is to be expected (Tilman et al., 1997; Lawton et al.,
Standardized values of mean phylogenetic diversity (MPDstand) and
1998; Prinzing et al., 2001). In our case, this relationship was
mean trait diversity (MTDstand) are based on comparisons of empirical significant, but rather weak, especially for epiphytes. Part of this
values with 1000 random assemblages, with two underlying random- mismatch can certainly be explained by the ‘diffuse nature’ of
ization procedures (unconstrained and constrained; see Materials and large datasets, especially the lack of finely resolved phylogenies.
Methods for details). Another explanation must also be taken into account: clades are

Journal of Biogeography 38, 394–405 399


ª 2010 Blackwell Publishing Ltd
J. Kluge and M. Kessler

(a)

(b)

Figure 3 Histograms of standardized values for mean phylogenetic diversity (MPDstand) and mean trait diversity (MTDstand) of species
within 156 fern assemblages along an elevational gradient in Costa Rica (a) for the unconstrained model and (b) for the constrained model.
A negative value indicates that phylogeny and traits of the species within the assemblage are more related than expected by chance
(clustering), while a positive value indicates that species are more diverse than expected by chance (overdispersion). Values of most
assemblages fall within the range [)2; 2], indicating that there are no significant deviations from 999 random assemblages and hence falling
within the 90% confidence interval of the models. Dark bars and dashed lines mark the limit of significant deviations. Applying correction
for multiple hypotheses testing (false discovery rate, see Materials and Methods) diminishes all cases of significant deviations.

phenotypically plastic, and similar traits may have evolved (Kluge et al., 2006). It has often been stated that elevational
independently in phylogenetically distant groups. This is transects form a natural gradient of rapidly changing environ-
especially interesting if phylogenetic and trait structures of mental shifts, especially in temperature and humidity regimes,
species assemblages show different trends along environmental and that particularly these factors govern species-richness
gradients. However, contrary observations exist as well. For patterns (Hawkins et al., 2003; Currie et al., 2004). Dispersal
example, de Bello et al. (2006) reported no consistent effect of limitation is not likely additionally to influence this pattern
increased richness on trait diversity in grassland assemblages because of the compact nature of our study area and the
along humidity and grazing gradients: while species richness efficient spore dispersal of ferns (Barrington, 1993). Thus the
varied considerably, the variety of traits did not differ between presence or absence of a phylogenetic fern group from a given
samples of high and low species richness. elevation along our study transect is determined rather by
The crucial question now is whether the assembly of species selection operating on the entire suite of physiological and
from a regional pool and their addition to the local pool are morphological traits of the taxa (Duckworth et al., 2000).
random processes, or whether species with a certain set of Values of the MPDstand and MTDstand structure of the fern
traits or from certain lineages are selectively favoured. assemblages were close to zero on average (Table 2), and after
correction for multiple hypothesis testing, none of the plots
showed significant deviations from random expectations
Structure of assemblages
(although individually considered plots did so; Fig. 3). How-
Within the study area, fern assemblages were outstandingly ever, this does not imply that no such patterns exist, as they
species-rich at mid-elevations and relatively species-poor at might occur among morphological or physiological traits not
both extremes of the elevational gradient, creating a symmet- studied by us, or with respect to ecological conditions we did
rical hump pattern in response to the environmental setting not include, such as microhabitat differentiation.

400 Journal of Biogeography 38, 394–405


ª 2010 Blackwell Publishing Ltd
Phylogenetic and trait diversity of ferns

(a)

(b)

Figure 4 Standardized values for mean phylogenetic diversity (MPDstand) and mean trait diversity (MTDstand) of 156 fern assemblages
based on comparisons of empirical MPD and MTD values with values from 999 random assemblages along an elevational gradient in Costa
Rica (a) for the unconstrained model and (b) for the constrained model. Each dot marks a plot; in contrast to Fig. 3, here we applied false
discovery rate to account for multiple hypotheses testing, leaving no individual plot with significant deviation from random assemblages.
However, trend lines across all data points combined were significant, and lines depict second-order polynomial regressions with *P < 0.05;
**P < 0.01; ***P < 0.001. Regressions and associated P values were corrected for spatial autocorrelation by applying simultaneous auto-
regressive models.

In any case, along the elevational gradient MPDstand and ful factors showed overdispersion and hence interspecific
MTDstand we documented various, but mainly very weak and competition.
above all insignificant, trends. The only exception, trait Although trait conservation has frequently been docu-
diversity patterns of epiphytes, raises two issues: the mismatch mented across most plant groups (Prinzing et al., 2001;
between elevational trends of phylogenetic and trait diversity Silvertown et al., 2006), the distribution of epiphytic taxa
in epiphytes, and the difference between both life forms in along this elevational gradient is close to random and constant,
general. thus being somewhat decoupled from the distribution of traits.
The humped trend of epiphytic trait richness may imply It seems that an overdispersed epiphytic trait structure at mid-
that the selection of species from the regional pool in these elevations is the product not only of specific processes of how
highly species-rich assemblages favours species that are more species may enter an assemblage, but also of how they might
distinct in terms of morphological features than a random evolve while persisting within the assemblage. As the taxa are
selection would be, and that are thus able to fill the available able to occur almost everywhere (random phylogenetic
microhabitats more effectively. As proposed by Kessler dispersion) and may be evolutionarily plastic, closely related
(2001), Hemp (2002), Bhattarai et al. (2004), Kluge et al. species coexisting at certain sites may undergo trait shifts
(2006) and others, the very humid conditions at mid- (‘character displacement’; Schluter, 2000) that reduce niche
elevations in tropical mountains represent optimal growing overlap. For example, Grant & Grant (2006) found this pattern
conditions for ferns, so climate is unlikely to act as a filter for finches on the Galápagos Islands, where closely related
excluding ill-adapted taxa. Under these conditions, compet- species rapidly developed different traits, thus enhancing trait
itive interactions may favour species with sets of traits richness while maintaining phylogenetic richness. Additionally,
different from those already represented. Similar results were the mid-elevations host not only the highest species richness,
recently found by Graham et al. (2009) for Andean hum- but also the highest number of individuals, suggesting that this
mingbird assemblages, where sites without ecologically stress- is the zone of highest productivity for ferns. It has been

Journal of Biogeography 38, 394–405 401


ª 2010 Blackwell Publishing Ltd
J. Kluge and M. Kessler

proposed that generation times are shorter in highly produc- deviations, especially along spatial and environmental gradi-
tive environments, resulting in higher diversification rates ents, gives insights into the processes governing species
(Cardillo, 1999). assembly. Our results were qualitatively congruent for both
Conversely, elevations with tendencies to trait clustering also null models and, as expected, the unconstrained null model
have low species richness. Here, species are apparently selected showed more pronounced trends due to more liberal condi-
for specific sets of traits that pass the filters dictated by the tions in constructing random assemblages.
local environmental conditions. Climatic conditions, distur-
bance regimes or biotic interactions have been suggested as the
ACKNOWLEDGEMENTS
main environmental filters removing species that lack appro-
priate traits for persistence under particular conditions This study was supported financially by grants from the
(‘assembly rules’ sensu Keddy, 1992; see also Dı́az et al., Deutsche Forschungsgemeinschaft (DFG) and Deutscher
1998; Duckworth et al., 2000). Likewise, convergent evolution Akademischer Austauschdienst (DAAD). We thank Mirkka
across the whole species pool may force species towards a Jones, Rafael Wüest and two anonymous referees for critical
suitable set of traits to cope with ambient conditions (e.g. comments and discussions on the manuscript. We thank
entire and stiff leaves to enhance frost or drought resistance; Sandrine Pavoine, Steven Kembel, Emmanuel Paradis, Karsten
Kluge & Kessler, 2007). Again, phylogenetic relatedness still Wesche, Henrik von Wehrden and Daniel Kissling for
may be random as these traits have evolved independently in statistical support and for advice concerning R packages.
different lineages (Cavender-Bares et al., 2004; Kraft et al.,
2007).
REFERENCES
In contrast to epiphytes, assemblages of terrestrial species
were found to stay almost random throughout the gradient. Ackerly, D.D. (2003) Community assembly, niche conserva-
Whatever the driving factor of assembly might be, the tism, and adaptive evolution in changing environments.
epiphytic life-form seems to be more subject to selecting International Journal of Plant Science, 164, 165–184.
processes. There are two possible explanations. First, the Barrington, D.S. (1993) Ecological and historical factors in fern
microhabitat diversity among height gradients within host biogeography. Journal of Biogeography, 20, 275–280.
trees (from tree base to outer canopy) is certainly higher by far de Bello, F., Lepš, J. & Sebastià, M.-.T. (2006) Variations in
than the soil variation within a 400 m2 plot, facilitating species and functional plant diversity along climatic and
differentiation within the epiphytic domain at mid-elevations. grazing gradients. Ecography, 29, 801–810.
Second, epiphytes are more exposed to, and thus coupled to, Benjamini, Y. & Yekutieli, D. (2001) The control of the false
environmental stress (Hietz & Hietz-Seifert, 1995) than discovery rate in multiple testing under dependency. Annals
terrestrial species, so that selecting forces may be more of Statistics, 29, 1165–1188.
pronounced. This interpretation is supported by the fact that Bhattarai, K.R., Vetaas, O.R. & Grytnes, J.A. (2004) Fern
species richness of epiphytes varies more strongly along the species richness along a central Himalayan elevational gra-
elevational gradient (from one to 48 species per plot) than that dient, Nepal. Journal of Biogeography, 31, 389–400.
of terrestrial species (three to 32), suggesting that epiphytic Bivand, R. (2006) spdep: Spatial dependence: weighting schemes,
fern assemblages are under relatively stronger limiting condi- statistics and models. R package version 0.4-59. Available at:
tions at the extremes of the gradient than terrestrials. http://CRAN.R-project.org/package=spdep.
Bivand, R.S., Pebesma, E.J. & Gómez-Rubio, V. (2008) Applied
spatial data analysis with R. Springer, Berlin.
CONCLUSIONS
Box, E.O. (1981) Macroclimate and plant growth forms: an
Our results show that both terrestrial and epiphytic fern introduction to predictive modeling in phytogeography. Junk,
assemblages exhibit almost random phylogenetic structure The Hague.
along the elevational gradient. Only epiphytic assemblages tend Cardelús, C., Colwell, R.K. & Watkins, J.E., Jr (2006) Vascular
towards trait overdispersion at mid-elevations and trait epiphyte distribution patterns: explaining the mid-elevation
clustering towards gradient extremes. This can be explained richness peak. Journal of Ecology, 94, 144–156.
by temperate and humid environmental conditions at mid- Cardillo, M. (1999) Latitude and rates of diversification in
elevations, which facilitate a wide trait range due to the lack of birds and butterflies. Proceedings of the Royal Society B:
strong environmental filters (frost, drought). Phylogenetic and Biological Sciences, 266, 1221–1225.
trait diversity do not show, at least for epiphytes, a strong Cavender-Bares, J., Ackerly, D.D., Baum, D.A. & Bazzaz, F.A.
relationship. This may reflect selective pressure that results in (2004) Phylogenetic overdispersion in Floridian oak com-
reduced trait and hence niche overlap between closely related munities. The American Naturalist, 163, 823–843.
species (‘selective trait shifts’). This process inflates trait space, Cavender-Bares, J., Kozak, K.H., Fine, P.V.A. & Kembel, S.W.
while the phylogenetic relationships are maintained. (2009) The merging of community ecology and phyloge-
More generally, we found that comparison of empirical and netic biology. Ecology Letters, 12, 693–715.
null assemblages enables the detection of assemblage structures Crozier, R.H. (1992) Genetic diversity and the agony of choice.
that deviate from random expectations. The direction of such Biological Conservation, 61, 11–15.

402 Journal of Biogeography 38, 394–405


ª 2010 Blackwell Publishing Ltd
Phylogenetic and trait diversity of ferns

Currie, D.J., Mittelbach, G.G., Cornell, H.V., Field, R., Guégan, Gotelli, N.J. & Graves, G.R. (1996) Null models in ecology.
J.-F., Hawkins, B.A., Kaufman, D.M., Kerr, J.T., Oberdorff, Smithsonian Institution Press, Washington, DC.
T., O’Brien, E. & Turner, J.R.G. (2004) Predictions and tests Gotelli, N.J. & Rohde, K. (2002) Co-occurrence of ectopara-
of climate-based hypotheses of broad-scale variation in sites of marine fishes: a null model analysis. Ecology Letters,
taxonomic richness. Ecology Letters, 7, 1121–1134. 5, 86–94.
Dabney, A., Storey, J.D. & Warnes, G.R. (2009) qvalue: Q-value Graham, C.H., Parra, J.L., Rahbek, C. & McGuire, J.A. (2009)
estimation for false discovery rate control. R package version Phylogenetic structure in tropical hummingbird communi-
1.20.0. Available at: http://CRAN.R-project.org/package ties. Proceedings of the National Academy of Sciences USA,
=qvalue. 106, 19673–19678.
Diáz, S. & Cabido, M. (1997) Plant functional types and eco- Grant, P.R. & Grant, B.R. (2006) Evolution of character dis-
system function in relation to global change. Journal of placement in Darwin’s finches. Science, 313, 224–226.
Vegetation Science, 8, 463–474. Grime, J.P. (2002) Plant strategies, vegetation processes, and
Dı́az, S. & Cabido, M. (2001) Vive la différence: plant func- ecosystem properties. J. Wiley & Sons, Chichester.
tional diversity matters to ecosystem processes. Trends in Halloy, S.R.P. & Mark, A.F. (1996) Comparative leaf mor-
Ecology and Evolution, 16, 646–655. phology spectra of plant communities in New Zealand, the
Dı́az, S., Cabido, M. & Casanoves, F. (1998) Plant functional Andes and the European Alps. Journal of the Royal Society of
traits and environmental filters at a regional scale. Journal of New Zealand, 26, 41–78.
Vegetation Science, 9, 113–122. Hawkins, B.A., Field, R., Cornell, H.V., Currie, D.J., Guégan,
Dormann, C.F., McPherson, J.M., Araújo, M.B., Bivand, R., J.-F., Kaufman, D.M., Kerr, J.T., Mittelbach, G.G., Obe-
Bolliger, J., Carl, G., Davies, R.G., Hirzel, A., Jetz, W., Kis- rdorff, T., O’Brien, E.M., Porter, E.E. & Turner, J.R.G.
sling, W.D., Kühn, I., Ohlemüller, R., Peres-Neto, P.R., (2003) Energy, water, and broad-scale geographic patterns
Reineking, B., Schröder, B., Schurr, F.M. & Wilson, R. of species richness. Ecology, 84, 3105–3117.
(2007) Methods to account for spatial autocorrelation in the Hemp, A. (2002) Ecology of the pteridophytes on the southern
analysis of species distributional data: a review. Ecography, slopes of Mt. Kilimanjaro I. Altitudinal distribution. Plant
30, 609–628. Ecology, 159, 211–239.
Dubuisson, J.Y., Hennequin, S., Rakotondrainibe, F. & Hietz, P. & Hietz-Seifert, U. (1995) Composition and ecology
Schneider, H. (2003) Ecological diversity in Trichomanes of vascular epiphyte communities along an altitudinal gra-
(Hymenophyllaceae). Botanical Journal of the Linnean Soci- dient in central Veracruz, Mexico. Journal of Vegetation
ety, 142, 41–63. Science, 6, 487–498.
Duckworth, J.C., Kent, M. & Ramsay, P.M. (2000) Plant Hooper, D.U., Chapin, S., Ewel, J., Hector, A., Inchausti, P.,
functional types: an alternative to taxonomic plant com- Lavorel, S., Lawton, J.H., Lodge, D.M., Loreau, M., Naeem,
munity description in biogeography? Progress in Physical S., Schmid, B., Setälä, H., Symstad, A.J., Vandermeer, J. &
Geography, 24, 515–542. Wardle, D.A. (2005) Effects of biodiversity on ecosystem
Ehleringer, J.R. & Mooney, H.A. (1978) Leaf hairs: effects on functioning: a consensus of current knowledge. Ecological
physiological activity and adaptive value to a desert shrub. Monographs, 75, 3–35.
Oecologia, 37, 183–200. Izsák, I. & Papp, L. (2000) A link between ecological diversity
Faith, D.P. (1992) Conservation evaluation and phylogenetic indices and measures of biodiversity. Ecological Modelling,
diversity. Biological Conservation, 61, 1–10. 130, 151–156.
Faith, D.P. (1995) Phylogenetic pattern and the quantification Janzen, D. (1985) On ecological fitting. Oikos, 45, 308–310.
of organismal biodiversity. Biodiversity, measurement and Keddy, P.A. (1992) Assembly and response rules: two goals for
estimation (ed. by D.L. Hawksworth), pp. 45–58. Chapman predictive community ecology. Journal of Vegetation Science,
& Hall, London. 3, 157–164.
Fargione, J., Tilman, D., Dybzinski, R., Hille Ris Lambers, J., Kembel, S., Ackerly, D., Blomberg, S., Cowan, P., Helmus, M.,
Clark, C., Harpole, W.S., Knops, J.M.H., Reich, P.B. & Lo- Morlon, H. & Webb, C. (2009) picante: R tools for integrating
reau, M. (2007) From selection to complementarity: shifts in phylogenies and ecology. R package version 0.7-1. Available
the causes of biodiversity–productivity relationships in a at: http://CRAN.R-project.org/package=picante.
long-term biodiversity experiment. Proceedings of the Royal Kembel, S.W. & Hubbell, S.P. (2006) The phylogenetic struc-
Society B: Biological Sciences, 274, 871–876. ture of a neotropical forest tree community. Ecology, 87,
Fridley, J.D. (2001) The influence of species diversity on eco- S86–S99.
system productivity: how, where, and why? Oikos, 93, 514–525. Kessler, M. (2001) Pteridophyte species richness in Andean
Gotelli, N.J. (2000) Null model analysis of species co-occur- forests in Bolivia. Biodiversity and Conservation, 10, 1473–
rence patterns. Ecology, 81, 2606–2621. 1495.
Gotelli, N.J. & Entsminger, G.L. (2001) Swap and fill algo- Kissling, W.D. & Carl, G. (2008) Spatial autocorrelation and
rithms in null model analysis: rethinking the knight’s tour. the selection of simultaneous autoregressive models. Global
Oecologia, 129, 281–291. Ecology and Biogeography, 17, 59–71.

Journal of Biogeography 38, 394–405 403


ª 2010 Blackwell Publishing Ltd
J. Kluge and M. Kessler

Kluge, J. & Kessler, M. (2007) Morphological characteristics of tical Computing, Vienna, Austria. Available at: http://
fern assemblages along an elevational gradient: patterns and www.R-project.org.
causes. Ecotropica, 13, 27–43. Ricotta, C. (2002) Measuring taxonomic diversity with para-
Kluge, J., Kessler, M. & Dunn, R.R. (2006) What drives ele- metric information functions. Community Ecology, 3, 95–99.
vational patterns of diversity? A test of geometric con- Ricotta, C. (2005) Through the jungle of biological diversity.
straints, climate and species pool effects for pteridophytes Acta Biotheoretica, 53, 29–38.
on an elevational gradient in Costa Rica. Global Ecology and Rojas, A. (1996) Twelve new species of Elaphoglossum
Biogeography, 15, 358–371. (Elaphoglossaceae) from Costa Rica and Panama. Brenesia,
Kluge, J., Bach, K. & Kessler, M. (2008) Elevational distribu- 45/46, 7–26.
tion and zonation of tropical pteridophyte assemblages in Rojas, A. (1997) Fourteen new species of Elaphoglossum
Costa Rica. Basic and Applied Ecology, 9, 35–43. (Elaphoglossaceae) from Mesoamerica. Brenesia, 47/48,
Koptur, S. & Lee, M.A.B. (1993) Plantlet formation in tropical 1–16.
montane ferns: a preliminary investigation. American Fern Rojas, A. (2001a) Seis especies nuevas y dos nuevos registros de
Journal, 83, 60–66. helechos (Pteridophyta) para Costa Rica. Revista de Biologia
Kraft, N.J.B., Cornwell, W.K., Webb, C.O. & Ackerly, D.D. Tropical, 49, 435–452.
(2007) Trait evolution, community assembly, and the phy- Rojas, A. (2001b) Nuevas especies, nombres nuevamente uti-
logenetic structure of ecological communities. The American lizados y nuevas distribuciones en los helechos arborescentes
Naturalist, 170, 271–283. (Filicales: Cyatheaceae) para el Neotrópico. Revista de Bio-
Kramer, U., Schneller, J.J. & Wollenweber, E. (1995) Farne und logia Tropical, 49, 453–466.
Farnverwandte. Georg-Thieme-Verlag, Stuttgart. Rojas, A. (2002a) New species, new combinations and new
Lavorel, S. & Garnier, E. (2002) Predicting changes in com- distributions in neotropical species of Elaphoglossum
munity composition and ecosystem functioning from plant (Lomariopsidaceae). Revista de Biologia Tropical, 50, 969–
traits: revisiting the Holy Grail. Functional Ecology, 16, 545– 1006.
556. Rojas, A. (2002b) Notes on Elaphoglossum (Lomariopsicaceae)
Lawton, J.H., Naeem, S., Thompson, L.J., Hector, A. & section Polytrichia subsection Hybrida in Mexico and Cen-
Crawley, M.J. (1998) Biodiversity and ecosystem function: tral America. Revista de Biologia Tropical, 51, 33–48.
getting the Ecotron experiment in its correct context. Sax, D.F., Stachowicz, J.J., Brown, J.H., Bruno, J.F., Dawson,
Functional Ecology, 12, 848–852. M.N., Gaines, S.D., Grosberg, R.K., Hastings, A., Holt, R.D.,
Legendre, P. (1993) Spatial autocorrelation: trouble or new Mayfield, M.M., O’Connor, M.I. & Rice, W.R. (2007) Eco-
paradigm? Ecology, 74, 1659–1673. logical and evolutionary insights from species invasions.
Lehsten, V. & Harmand, P. (2006) Null models for species co- Trends in Ecology and Evolution, 22, 465–471.
occurrence patterns: assessing bias and minimum iteration Schluter, D. (2000) Ecological character displacement in
number for the sequential swap. Ecography, 29, 786–792. adaptive radiation. The American Naturalist, 156, S4–S16.
Lomolino, M.V., Riddle, B.R. & Brown, J.H. (2006) Biogeog- Schweiger, O., Klotz, S., Durka, W. & Kühn, I. (2008) A
raphy. Sinauer, Sunderland, MA. comparative test of phylogenetic diversity indices. Oecologia,
MacArthur, R.H. (1958) Population ecology of some warblers 157, 485–495.
of northeastern coniferous forests. Ecology, 39, 599–619. Silvertown, J., Dodd, M., Gowing, D., Lawson, C. & McCon-
Magurran, A.E. (1988) Ecological diversity and its measure- way, K. (2006) Phylogeny and the hierarchical organization
ments. Princeton University Press, Princeton, NJ. of plant diversity. Ecology, 87, S39–S49.
Mickel, J.T. (1976) Vegetative propagation in Asplenium Smith, B., Moore, S.H., Grove, P.B., Harris, N.S., Mann, S. &
exiguum. American Fern Journal, 66, 81–82. Wilson, J.B. (1994) Vegetation texture as an approach to
Moran, R.C. & Riba, R. (1995) Flora Mesoamericana. Vol. 1. community structure – community-level convergence in a
Psilotaceae a Salviniaceae. Universidad Nacional Autónoma New Zealand temperate rainforest. New Zealand Journal of
de Mexico, Mexico. Ecology, 18, 41–50.
Moulton, M.P. & Pimm, S.L. (1987) Morphological assort- Smith, W.K. & McClean, T.M. (1989) Adaptive relations
ment in introduced Hawaiian passerines. Evolutionary between leaf water repellency, stomatal distribution, and gas
Ecology, 1, 113–124. exchange. American Journal of Botany, 76, 465–469.
Mouquet, N., Moore, J.L. & Loreau, M. (2002) Plant species Storey, J.D. (2002) A direct approach to false discovery rates.
richness and community productivity: why the mechanism Journal of the Royal Statistical Society. Series B, Statistical
that promotes coexistence matters. Ecology Letters, 5, 56–65. Methodology, 64, 479–498.
Prinzing, A., Durka, W., Klotz, S. & Brandl, R. (2001) The Stubbs, W.J. & Wilson, J.B. (2004) Evidence for limiting
niche of higher plants: evidence for phylogenetic conserva- similarity in a sand dune community. Journal of Ecology, 92,
tism. Proceedings of the Royal Society B: Biological Sciences, 557–567.
268, 2383–2389. Swenson, N.G. (2009) Phylogenetic resolution and quantifying
R Development Core Team (2009) R: a language and envi- the phylogenetic diversity and dispersion of communities.
ronment for statistical computing. R Foundation for Statis- PLoS ONE, 4, e4390.

404 Journal of Biogeography 38, 394–405


ª 2010 Blackwell Publishing Ltd
Phylogenetic and trait diversity of ferns

Tilman, D., Knops, J., Wedin, D., Reich, P., Ritchie, M. & Appendix S1 Standardized regression coefficients of ordin-
Siemann, E. (1997) The influence of functional diversity and ary least squares and simultaneous autoregressive models of
composition on ecosystem processes. Science, 277, 1300– fern richness distribution along an elevational gradient in
1302. Costa Rica.
Vane-Wright, R.I., Humphries, C.J. & Williams, P.H. (1991)
As a service to our authors and readers, this journal provides
What to protect? – Systematics and the agony of choice.
supporting information supplied by the authors. Such mate-
Biological Conservation, 55, 235–254.
rials are peer-reviewed and may be reorganized for online
Watkins, J.E., Jr, Cardelús, C., Colwell, R.K. & Moran, R.C.
delivery, but are not copy-edited or typeset. Technical support
(2006) Species richness and distribution of ferns along an
issues arising from supporting information (other than
elevational gradient in Costa Rica. American Journal of
missing files) should be addressed to the authors.
Botany, 93, 73–83.
Webb, C.O. (2000) Exploring the phylogenetic structure of
ecological communities: an example for rain forest trees. The
American Naturalist, 156, 145–155.
Webb, C.O., Ackerly, D.D., McPeek, M.A. & Donoghue, M.J.
(2002) Phylogenies and community ecology. Annual Review
of Ecology and Systematics, 33, 475–505. BIOSKETCHES
Weiher, E. & Keddy, P.A. (1995) Assembly rules, null models,
and trait dispersion: new questions from old patterns. Oikos, Jürgen Kluge is a lecturer at the Faculty of Geography at the
74, 159–164. University of Marburg, and research assistant at the Depart-
Wiens, J.J. & Graham, C.H. (2005) Niche conservativism: ment of Systematic Botany of the University of Zurich. He has
integrating evolution, ecology, and conservation biology. research interests in the field of diversity and biogeography
Annual Review of Ecology, Evolution, and Systematics, 36, with a special focus on pteridophytes and bryophytes.
519–539. Michael Kessler is researcher at the Department of System-
Yachi, S. & Loreau, M. (2007) Does complementary resource use atic Botany of the University of Zurich, and scientific curator
enhance ecosystem functioning? A model of light competition of the Botanical Garden of the university. His research focuses
in plant communities. Ecology Letters, 10, 54–62. on the systematics, diversity, biogeography and conservation
of tropical plants and birds, with a special focus on mountain
SUPPORTING INFORMATION forests.

Additional Supporting Information may be found in the


online version of this article: Editor: Ole Vetaas

Journal of Biogeography 38, 394–405 405


ª 2010 Blackwell Publishing Ltd

You might also like