Oikos 116: 882
892, 2007

doi: 10.1111/j.2007.0030-1299.15559.x,
Copyright # Oikos 2007, ISSN 0030-1299
Subject Editor: Pia Mutikainen, Accepted 8 January 2007

Let the concept of trait be functional!

Cyrille Violle, Marie-Laure Navas, Denis Vile, Elena Kazakou, Claire Fortunel,
Irène Hummel and Eric Garnier
Cyrille Violle ([email protected]), Marie-Laure Navas, Denis Vile, Elena Kazakou, Claire Fortunel, Ire`ne Hummel and Eric
Garnier, CNRS, Centre d’Ecologie Fonctionnelle et Evolutive, UMR 5175, 1919, Route de Mende, FR-34293 Montpellier Cedex 5,
France.
MLN also at: Montpellier Supagro, 2 Place Viala, FR-34060 Montpellier Cedex 1, France. DV also at: De´pt de Biologie,
Univ. de Sherbrooke, Sherbrooke (QC), Canada, J1K2R1.

In its simplest definition, a trait is a surrogate of organismal performance, and this meaning of the term has been
used by evolutionists for a long time. Over the last three decades, developments in community and ecosystem
ecology have forced the concept of trait beyond these original boundaries, and trait-based approaches are now
widely used in studies ranging from the level of organisms to that of ecosystems. Despite some attempts to fix
the terminology, especially in plant ecology, there is currently a high degree of confusion in the use, not only of
the term ‘‘trait’’ itself, but also in the underlying concepts it refers to. We therefore give an unambiguous
definition of plant trait, with a particular emphasis on functional trait. A hierarchical perspective is proposed,
extending the ‘‘performance paradigm’’ to plant ecology. ‘‘Functional traits’’ are defined as morpho-physio-
phenological traits which impact fitness indirectly via their effects on growth, reproduction and survival, the
three components of individual performance. We finally present an integrative framework explaining how
changes in trait values due to environmental variations are translated into organismal performance, and how
these changes may influence processes at higher organizational levels. We argue that this can be achieved by
developing ‘‘integration functions’’ which can be grouped into functional response (community level) and effect
(ecosystem level) algorithms.

Trait-based approaches are widely used in ecological defined at higher organizational levels (Fig. 1). Despite
and evolutionary research. Historically, the term ‘‘trait’’ some attempts to fix the terminology (Lavorel et al.
has naturally moved from the common language to a 1997, Semenova and van der Maarel 2000, Dı́az and
more scientific one in different disciplines (e.g. quanti- Cabido 2001), there is currently a high degree of
tative genetics, physiological ecology, functional ecol- confusion in the use, not only of the term ‘‘trait’’ itself,
ogy, population demography, evolutionary physiology, but also in the underlying concepts it refers to.
life-history evolution). Following Darwin’s (1859) A further complication has arisen with the introduc-
proposal, traits were initially mainly used as predictors tion of ‘‘functional traits’’, an expression coined with
(proxies) of organismal performance. Over the last three the emergence of functional ecology as a discipline
decades, developments in community (Grime 1974, (Calow 1987, Keddy 1992a). Current attempts to
Petchey and Gaston 2002, McGill et al. 2006) and devise functional classifications of organisms (Noble
ecosystem (Chapin 1993, Grime 1998, Lavorel and and Gitay 1996, Woodward and Cramer 1996, Lavorel
Garnier 2002, Eviner and Chapin 2003) ecology have et al. 1997), define the functional diversity of commu-
forced the concept of trait beyond these original nities (Walker et al. 1999, Petchey and Gaston 2002,
boundaries, and trait-based approaches are now used Mason et al. 2005) and assess species effects on
in studies ranging from the level of organisms to that of ecosystems (Lavorel and Garnier 2002, Eviner and
ecosystems. Within each discipline, diverse types of Chapin 2003), all advocate the use of ‘‘functional
traits are thus used to assess inherent components traits’’. McGill et al. (2006) even proposed that
involved in the explanation of complex processes functional traits should be the corner stone in a

882
 Leveloforganization
Level of organization Challenge of
ofinterest
interest Inherentcomponents
Inherent components Examplesoftraits
Examples of traits

Growth Ecophysiological traits

Individual fitness
Individual Fecundity
w
Survival Life-history traits

II-P

Birth
Population Finite rate of increase Death
Demographic traits
λ
? Immigration
Emigration

IP-C

Community structure Occurrence

Community Response traits
and dynamics Frequency

IC-E

Ecosystem Ecosystem properties

Ecosystem (energy flow and Effect traits
functioning chemical cycling)

Fig. 1. Pathways linking the challenge of interest of different organizational levels, through their related inherent components, to
some examples of traits found in the literature. Without trait-based information, scaling-up to higher organizational levels needs
complex integration information (I). Thus fitness components of an individual determine the components of the finite rate of
increase (l) of the population (II
P). Occurrence and frequency of species at the community level encompass components of l
through complex integration (e.g. biotic interactions) (IP
C). Finally, scaling-up to ecosystem properties can be done by
combining functional property of each species of the community (IC
E). Using traits as proxies of a process at a particular
organizational level can sometimes be done without such integration function. For example, at the ecosystem level, ecosystem
productivity (one component of ecosystem functioning) shows a strong positive relationship with plant height (an effect trait)
(Saugier et al. 2001).

rebuilding of community ecology. However, as it is the level (four leaf traits, three traits describing the stature
case for ‘‘traits’’, the definition and use of ‘‘functional of the plants and seed mass), while three were measured
traits’’ remains ambiguous, as we will show below. at a plot level (standing biomass, vegetation cover and
The aims of this paper are to propose unambiguous canopy height). In the second one, Eviner (2004)
definitions of ‘‘traits’’ and ‘‘functional traits,’’ and to characterized the influence of plant standing biomass,
suggest a framework on how traits could be used to soil conditions and microbial phosphorus, all defined as
address current issues in community and ecosystem ‘‘plant traits’’ (see Table 4 of Eviner’s paper for details),
ecology. Although some concepts and examples will be on ecosystem properties such as litter characteristics,
drawn from animals, the main focus will be on plants. labile carbon inputs, and on soil temperature and
moisture. In both studies, the dependent variables
were defined at the ecosystem level, and were explained
Plant traits: definitions by some ‘‘traits’’, considered as independent variables.
However, as the lists above show, plant ‘‘traits’’
Two recent papers dealing with the impact of plant characterized the features of individuals, whole com-
functional diversity on ecosystem properties are repre- munities and even soil conditions. Lumping such
sentative of the degree of confusion in the use of the different features under the same word ‘‘trait’’ therefore
term ‘‘trait’’. In the first one, Petchey et al. (2004) mixes very different components of community struc-
tested how different indices of functional diversity ture: the number of individuals per unit ground area
could explain differences in biomass production. These and the characteristics of these individuals (see ‘‘Inte-
indices were calculated using a number of so-called gration functions’’ below) and, in the case of soil
‘‘traits’’: eight were measured at the individual plant conditions, their environment. Obviously, the factors

883
influencing e.g. the number of individuals are likely to is required to define a trait. At the population level, we
be somewhat different from those influencing their suggest to stick to the terminology used by e.g. Caswell
characteristics. We argue that using the same term (1989) and use ‘‘demographic parameters’’ instead of
‘‘trait’’ to designate such different aspects of a commu- traits. At the community or ecosystem level, we suggest
nity can only lead to an unproductive confusion in the to use the term ‘‘property’’ to designate any feature or
identification of the underlying mechanisms controlling process such as community diversity, standing biomass,
community structure and ecosystem functioning. Kill- decomposition or soil resource availabilities.
ingbeck (1986) argues for four criteria to validate an This definition of a trait requires further precisions:
appropriate term: (1) unambiguous, (2) descriptive, (3) (1) the particular value or modality taken by the trait at
as simple and short as possible, and (4) frequently used any place and time is called an ‘‘attribute’’ (Lavorel et al.
to describe the [feature] in question. These examples 1997); (2) within a species, the trait, either continuous
show that, as currently used, ‘‘trait’’ apparently con- or categorical, may show different attributes along
forms to criteria (3) only. environmental gradients (Fig. 2) or through time; (3)
Given the wide range of research fields that use the the attribute for a trait is usually assessed for one
term ‘‘trait’’ (Fig. 1), suggesting a proper definition population (average of attributes of a set of individuals)
appears far from trivial. In many cases, a trait is defined in space and time (i.e. for a single value of the x-axis of
at the level of the individual (Table 1, 2). Noticeable Fig. 2). This has two consequences. First, there is not a
exceptions include traits defined at the plot/ecosystem single trait attribute for any particular species (Fig. 2).
level as discussed above, and ‘‘demographic traits’’ Assessing intra- vs inter-specific variability of traits is
defined at the population level. To avoid confusion, clearly of interest in this context (Garnier et al. 2001,
we suggest that traits be used at the level of the Cornelissen et al. 2003, Roche et al. 2004). Even
individual only, with the following definition: a trait is categorical traits such as life-form or the ability to fix
any morphological, physiological or phenological fea- atmospheric nitrogen, which are generally considered as
ture measurable at the individual level, from the cell to inherent to the species, may actually vary with environ-
the whole-organism level, without reference to the mental conditions: for example, it is well known that
environment or any other level of organization. This nitrogen fixation is inhibited at high soil nitrogen
definition implies that no information external to the availability (Hartwig 1998). Second, information on
individual (environmental factors) or at any other level the local environment where the trait has been measured
of organization (population, community or ecosystem) is essential to interpret the ecological/evolutionary

Table 1. Examples of traits used in the literature, with the function, component or process they are supposed to capture the levels of
definition and application, and selected references.

Trait What does the trait describe? Level of Level of Examples

definition application

Ecophysiological trait Relative fitness Individual Individual Resource uptake and utilization
(e.g. leaf size, photosynthetic rate)
(Ackerly et al. 2000)
Functional trait See Table 2
Life-history trait (1) Relative fitness Individual Individual Insect adult body (Honek 1993)
(2) Phenology Individual Individual Lifespan, time to first reproduction
(McIntyre et al. 1999)
(3) Demographic traits and finite Individual Population Clutch size, age at maturity in bird
rate of increase ecology (Saether and Bakke 2000)
of a population (l)
Demographic trait* Finite rate of increase of a Population Population Rates of birth, death, immigration
population (l) and emigration (Caswell 1989)
Vital attribute Vegetation replacement sequences Individual Community Ability to establish and grow to
(Noble and Slatyer 1980) maturity in a community (Noble
and Slatyer 1980)
Biological trait Species distribution (survival Individual Community e.g. growth form, leaf area, fruit
ability in an environment) size for distribution of hydrophytes
(Willby et al. 2000)

* many views of ‘‘demographic traits’’ have been found in the population ecology literature. We restrict the term ‘‘demographic
traits’’ to population-level measures which directly condition the finite rate of increase (l): age- or stage-specific rates of survival,
reproduction, growth, development, etc (‘‘vital rates’’ in Caswell 1989, 1996). One of the aims of population dynamics is to explore
the sensibility of the population rate of increase to changes in any of the vital rates through changes in organismal traits which are
called ‘‘life-history traits’’ (or sometimes ‘‘demographic traits’’ sensu lato). With this restricted definition, the level of definition of
demographic traits is only the population one

884
Table 2. Examples of plant functional traits found in literature, with the function, component or process they are supposed to
capture the levels of definition and application.

Trait What does the trait describe? Level of Level of

definition application

Functional trait (1) Physiological process (e.g. photosynthesis, respiration) Individual Individual
(Calow 1987) (physiological view)
(2) Life-history process (e.g. germination, growth, reproduction) Individual Individual/
(Weiher et al. 1999) (demographic view) population
(3) Individual fitness (Reich et al. 2003) (fitness-related view) Individual Individual
(4) Performance measures (Geber and Griffen 2003, Individual Individual
McGill et al. 2006) (performance-related view)
Performance trait Fitness components (growth reproduction, survival) (Geber and Individual Individual
Griffen 2003)
Response trait Response of a plant to environmental changes (Keddy 1992b, Individual Individual
Lavorel and Garnier 2002)
Effect trait Effect of a plant on ecosystem functioning (Dı́az and Individual Ecosystem
Cabido 2001, Lavorel and Garnier 2002)
Hard trait Function s.l. at any organizational level (Hodgson et al. 1999, Individual Any organizational
Weiher et al. 1999) level
Soft trait Hard trait (soft trait is more easily measurable than the hard Individual Individual
trait itself) (Hodgson et al. 1999, Weiher et al. 1999)
Functional marker Function s.l. at any organizational level (more easily Individual Any organizational
measurable than the function itself) (Garnier et al. 2004) level

A meaning of trait attributes (McGill et al. 2006), even if

this information is not compulsory to define a trait. This
information is provided in recent data bases (Knevel
et al. 2005, Garnier et al. 2006) which, in addition to
the trait mean and variance, provide quantitative data
T rait

for some environmental variables (e.g. climate, soil

characteristics, disturbance regimes) of the habitat where
the population selected for trait measurement has been
sampled.

A hierarchical perspective on plant traits

B Arnold’s framework applied to plants
1
The performance of a species in a particular ecological
habitat is assessed by the ability of a sample of
individuals to grow, reproduce or survive (fitness
T rait

components in Fig. 1). In animals, this ability or

Darwinian fitness is assessed by measuring components
of whole-organism performance such as body size or age
at maturity (Le Galliard et al. 2004). Such ‘‘perfor-
mance traits’’ are a key concept of the ‘‘morphology,
0 performance, and fitness’’ paradigm proposed by
Arnold (1983) for animals. According to this paradigm,
Environmental variable morphological traits influence (directly and indirectly)
performance traits, which in turn influence (directly or
Fig. 2. Theoretical examples of intraspecific variability of a indirectly) fitness. For example, when evaluating the
trait (continuous: A, categorical: B) in response to environ-
locomotor performance in lizards, sprint performance
mental changes. Each point corresponds to the trait value
taken by one individual at a point of an environmental (a performance trait) can have a direct impact on fitness
gradient. The dashed line represents an example of fitted (i.e. increasing survival by escaping from predators), yet
functions; in the case of categorical trait (B), logistic is the net result of many morpho-physiological traits,
regressions can be performed (here a quadratic function has including limb morphology, muscle fibre composition
been used). or neuronal capabilities (Wikelski and Romero 2003).

885
The ‘‘performance paradigm’’ constitutes a very useful vegetative biomass at the end of a growing season.
framework, which has been introduced only very Recent advances in plant ecophysiology have gone one
recently in plant ecology (Ackerly et al. 2000, Geber step forward, and identified the underlying M-P-P traits
and Griffen 2003). As argued by McGill et al. (2006) which had the strongest impacts on RGR (Poorter and
who advocate the use of unified ‘‘performance curren- van der Werf 1998, Shipley 2006), one of them being
cies’’ to improve our understanding of plant commu- specific leaf area (cf. Fig. 4B). The M-P-P traits may be
nity, acknowledging the importance of individual inter-related (Fig. 3): among a wealth of examples, one
performance is central in ecology. We suggest that the such relationship is that between specific leaf area (the
three components of plant fitness (growth, reproduc- ratio of leaf area to leaf dry mass) and net photosynthesic
tion and survival) be assessed by the measure of only rate observed both at the leaf (Fig. 4A) and whole-plant
three performance traits: vegetative biomass, reproduc- (Fig. 4B) levels. The search for relationships between
tive output and plant survival (Fig. 3). Vegetative M-P-P traits and performance traits is an on-going quest
biomass represents the net cumulated outcome of all in plant ecology (Ackerly et al. 2000, Geber and Griffen
growth and loss processes, reproductive output can be 2003), at the cross-roads of ecophysiology, ecological
measured by the seed biomass plus the reproductive genetics and community ecology.
accessories and/or the number of seeds produced, and The identification of performance traits imposes a
plant survival is assessed by a binary variable (an hierarchy among plant traits, which can be important in
individual is dead or alive in the environmental statistical analysis (Arnold 1983), using for example
condition considered) or a probability (in this case, structural equation modelling (Ackerly et al. 2000, Vile
information about a cohort of individuals is needed). et al. 2006a). More generally, it may prove useful to
The value of these three performance traits is influenced address particular questions such as the identification of
by morphological, physiological and phenological suites of quantitative traits to explain ecosystem
(M-P-P) traits, operating from the cell to the whole functioning (Eviner and Chapin 2003) or plant
plant levels (Fig. 3). strategies (Vile et al. 2006a), or to assess functional
In the case of vegetative biomass, a convenient diversity based on the clustering of plant traits (Petchey
framework may be used to unravel relationships and Gaston 2002).
between M-P-P traits and performance traits. Vegeta-
tive biomass at any time t (VMt) can be written as:
The special case of functional traits
VMt VM0 eRGRt (1)
‘‘Plant functional trait’’ is a currently widely used
this simple equation shows that VM0, the initial plant expression in plant ecology (Dı́az and Cabido 2001,
mass (e.g. seed mass in the case of annuals), RGR, the Lavorel and Garnier 2002, Hooper et al. 2005), but its
relative growth rate, and t, the duration of active growth, actual meaning varies substantially among authors
potentially play a role in the determination of the (Table 2). It may be understood as a surrogate of a

M-P-P trait 1

M-P-P trait 2
Vegetative biomass

M-P-P trait 3
Reproductive output Plant performance Individual fitness
M-P-P trait 4

Plant survival
...

M-P-P trait k

Functional traits Performance traits

Traits

Fig. 3. Arnold’s (1983) framework revisited in a plant ecology perspective. Morpho-physio-phenological (M-P-P) traits (from 1
to k) modulate one or all three performance traits (vegetative biomass, reproductive output and plant survival) which determine
plant performance and, in fine, its individual fitness. M-P-P traits may be inter-related (dashed double-arrows). For clarity, inter-
relations among performance traits and feedbacks between performance and M-P-P traits are not represented.

886
 1000
A
function (e.g. specific leaf area: see above) or as this
s )
-1 -1 leaf
function itself (e.g. photosynthesis) (Fig. 4A), with the
difficulty to agree on the actual meaning of function
Net photosyntetic rate (nmol CO2 g

(Calow 1987, Jax 2005; Table 2). It may also be

100
considered as a trait that strongly influences organismal
performance (McGill et al. 2006) and/or individual
fitness (Geber and Griffen 2003, Reich et al. 2003).
Finally, it may be defined with respect to ecosystem
10 functioning (McIntyre et al. 1999): this is the case of
functional effect traits, defined as those traits that have
r = 0.71*** an impact on ecosystem functioning (Dı́az and Cabido
(n = 764)
2001, Lavorel and Garnier 2002). To simplify the
1 concept, we propose to define a functional trait as any
1 10 100
M-P-P trait which impacts fitness indirectly via its
0.4 effects on performance traits (Fig. 3).
whole plant
B Hodgson et al. (1999) and Weiher et al. (1999) have
introduced the hard/soft dichotomy in the functional
d )
-1 -1

0.3 trait terminology (Table 2). ‘‘Hard’’ traits are those

which capture the function of interest, but which are
Relative growth rate (g g

either difficult or expensive to measure, while soft traits

0.2 are surrogates of such functions, but are less difficult
and/or expensive to obtain. However this terminology
is rather subjective and hardly operational. For example,
0.1 whether a trait is expensive is relative to the availability
and salary of people and/or equipment to make the
r = 0.71***
(n = 272)
measurement, while easiness of measurement is often
0.0 only apparent. A good example is the case of plant
0 10 20 30 40 50 60 height, usually considered as a surrogate of competitive
ability. A recent study has clearly shown that for
10
C herbaceous species, this assumption is only valid for a
Specific above-ground net primary productivity

community certain period of time during growth (Violle et al.,

8 unpubl.), which is not particularly easy to assess. As a
consequence, we suggest to preclude the use of the
6
hard/soft trait terminology.
(g kg d )
-1 -1

4
Plant traits and environment
2 Environmental factors can be considered as filters
r = 0.78**
(n = 12)
constraining which individuals bearing specific attri-
butes of ‘‘response traits’’ are able to be sorted out and
0
10 15 20 25 30 persist in a community (Keddy 1992b). Different sets
2 -1
Specific leaf area (m kg )
of response traits to environmental factors such as
resources and disturbances have been recognized in
Fig. 4. Relationships between specific leaf area and processes plants (Chapin et al. 1993, Grime et al. 1997, Lavorel
measured at three different organization levels. (A): leaf level and Garnier 2002, Ackerly 2004). For example, age at
(drawn from data taken in Wright et al. 2004); (B): whole maturity, seedling relative growth rate, growth form,
plant level (drawn from data taken in Wright and Westoby
shoot height, are response traits to fire (Lavorel and
2001); (C): community level (drawn from data taken in
Garnier et al. 2004); in this latter case, specific leaf area is Garnier 2002).
weighed by the abundance of species in the communities Following animal ecologists (Irschick and Garland
(community functional parameter). Pearson correlation coef- 2001, Irschick 2003), we propose that the response of
ficients (r) and number of data points (between brackets) are the whole-organism performance to an environmental
given for each relationship. Significance levels: **, p B/0.01; variable be called an ‘‘ecological performance’’. Thus an
***, pB/0.001. Note that the quantities on both axes are ecological performance can be defined as the optimum
equivalent and expressed in comparable units across the three and/or the breadth of distribution of performance traits
organizational levels. along an environmental gradient (Fig. 5). Examples

887
 Optimum individuals per unit ground area) by information at the
individual level (plant biomass, one of the three
performance traits as discussed above). More complex
Performance trait

Ecological perfomance
integration functions have been proposed and used to
scale-up from plant traits to communities (McGill et al.
2006) and ecosystems (Garnier et al. 2004, Kerkhoff
Breadth
and Enquist 2006), and their precise form will depend
on the property under study. Below, a general frame-
work showing how this scaling-up could be done is
Environmental variable presented, using ‘‘functional response and effect algo-
rithms’’ as proposed by Naeem and Wright (2003).
Fig. 5. Graphical definition of an ecological performance. An From a theoretical perspective, understanding and
ecological performance can be defined as the optimum and/or
predicting the functioning of an ecosystem (the highest
the breadth of distribution of one of the three performance
traits along an environmental gradient; see text for more organization level described in Fig. 1) requires the
details. knowledge of the identity of species, the number of
individuals of each species and of the attributes of these
species (Fig. 1). A first step is therefore to assess the
include environmental tolerances (e.g. to shade,
relative abundance of species i in a community j (Ai,j),
drought, heavy-metals, or herbivory), habitat or ecolo-
which is a function of the demographic parameters
gical preferences (e.g. Ellenberg’s numbers, Ellenberg
(DPi,j) of each population composing this community,
1988). These are sometimes called ‘‘traits’’ (e.g. shade
and the performance traits (PTi,j) of the individuals
tolerance, Huston and Smith 1987, frost tolerance-
from these populations. This can be written as:
McGill et al. 2006, habitat or ecological preferences by
species abundance, Pywell et al. 2003), but according to
the definition given above, this should not be so, since
external (here environmental) variables are required for
Aij 
g f (PT ; DP ) dt
t
ij ij (3)

their definition. Actually, ecological performances de-

pend on the coordinated response of multiple traits to Time-integration accounts for the age-structure of each
environmental factors, in agreement with the hierarch- population, which is of interest when searching to
ical scheme proposed above. For example, grazing explain changes in plant community composition. Pros
tolerance is related to a variety of attributes such as and cons of using demographic parameters in commu-
short lifespan, short and prostate stature, stoloniferus nity dynamics is currently lively debated since these are
and rosette architecture (Dı́az et al. 2006). difficult to measure in the field and can increase
dramatically with the number of species concerned
(McGill et al. 2006). However no simple substitute is
available, and this is undoubtedly an area where
Scaling-up from organisms to population, conceptual and methodological breakthroughs are
community or ecosystem urgently needed (see for example Shipley et al. 2006,
for a recent attempt to bypass the complexity of species
Scaling-up from organisms to higher organizational demography and assembly rules using functional traits
levels is a process of major interest in ecology (Field and as defined above).
Ehleringer 1993, Duarte et al. 1995, Jones and Lawton Although undoubtedly a long-term goal in ecology,
1995), but this is probably an area where the use of the the mapping from performance traits to population
plant trait terminology is most confusing. We argue dynamics and community structure (IIP and IPC
that scaling from plant traits to populations and/or integration functions in Fig. 1) is currently poorly
communities and/or ecosystems requires explicit ‘‘in- understood (McGill et al. 2006). Another approach has
tegration functions’’ (Fig. 1). They can be very simple been to assess the effects of species on ecosystem
when effects of individuals are additive. For example, functioning, without taking community dynamics into
the standing biomass of a monospecific community is: account. This has led to the formulation of the ‘‘biomass
ratio hypothesis’’ (Grime 1998), which states that
Standing biomassi (g m2 ) particular traits of locally abundant species will deter-
 Ni (nbind m2 )Bi (g ind1 ) (2) mine the rate and magnitude of instantaneous ecosystem
processes. Such traits have been called ‘‘effect traits’’
where Ni is the number of individuals (nbind) in (Dı́az and Cabido 2001, Lavorel and Garnier 2002). A
community i, and Bi the average biomass of the limited number of tests have shown that this hypothesis
individuals. In this example, the integration is made held for some key biogeochemical processes such as
by combining a community-level property (number of specific primary productivity and litter decomposition

888
(Garnier et al. 2004, Vile et al. 2006b, Quétier et al. (e.g. plant-soil feedbacks, complex biotic interactions,
unpubl.). Ecosystem properties therefore depend on functional complementarity; Dı́az et al. 2007).
effect traits (ET) weighted according to the relative Acknowledging that species traits have potential
abundance of species. The term ‘‘community-aggregated impacts at higher organization levels protracts the
trait’’ has been proposed to define this community-level concept of ‘‘extended phenotype’’
i.e. the effects of
feature, but this is not in agreement with our definition genes at levels higher than the population
developed
of trait, which should be restricted to the individual in a perspective of a community and ecosystem genetics
level. Therefore, by analogy with the ‘‘demographic (Whitham et al. 2003). The framework of integration
parameter’’ which is a population-level feature based on functions proposed here at the species level can easily be
aggregation of individual level features, we propose the extended to any group of individuals sharing the same
term ‘‘community functional parameter’’ (CFP) for traits (e.g. genotypes, populations, functional groups)
such a community-aggregated feature: or being at the same demographic or ontogenic stages.
X
nj In this case, the probability distributions of traits in the
CFPj  Ak;j ETk;j (4) community are required. The main advantage is that
k1 this integration might be done without taxinomic or
genetic information and that it takes into account inter-
where nj is the number of species sampled in community
individual variance, an important aspect of functional
j, Ak,j is the relative abundance of species k in
community j, and ETk,j is the effect trait of species k diversity (Petchey and Gaston 2006). Overall, this
in community j. For example, community-aggregated suggests that approaches recently developed in commu-
specific leaf area has been shown to be related to specific nity and ecosystem ecology apply to the emerging field
above-ground net primary productivity of herbaceous of community/ecosystem genetics.
communities (Fig. 4C). This shows that information on
community structure (local abundances of species) and
functional identity of species (specific leaf area) can be Conclusions
combined into integration functions (IC
E in Fig. 1) to
explain the functioning of ecosystems of known com- Plant traits are increasingly used to address questions
position. Integration function might be non-linear when from organisms to ecosystems and beyond. For the con-
the effects of species composition on community cept to be functional, we suggest to stick to the termino-
dynamics and ecosystem functioning are not additive logy summarized in Table 3. The main points are:

Table 3. Summary of definitions proposed and discussed in this paper.

Terms Definitions

Individual level:
Overall definitions:
“/ Trait “/ Any morphological, physiological or phenological feature measurable at the
individual level, from the cell to the whole-organism level, without reference
to the environment or any other level of organization.
“/ Attribute “/ Value or modality taken by a trait at a point of an environmental gradient.
“/ Functional trait “/ Any trait which impacts fitness indirectly via its effects on growth,
reproduction and survival.
“/ Performance trait “/ Direct measure of fitness. In plants, only three types of performance traits are
recognized: vegetative biomass, reproductive output (e.g. seed biomass, seed
number), plant survival.
Interactions with environment:
“/ Response trait “/ Any trait the attribute of which varies in response to changes in environmental
conditions.
“/ Ecological performance “/ Response of the whole-organism performance, assessed by one or more
performance traits (maximum, mean or variance), to an environmental
gradient.
“/ Effect trait “/ Any trait which reflects the effects of a plant on environmental conditions;
community or ecosystem properties.
Population, community and ecosystem levels:
“/ Demographic parameter “/ Population feature which directly conditions the finite rate of increase (l) of the
population: age- or stage-specific rates of survival, reproduction, growth,
development.
“/ Community or ecosystem property “/ Any feature or process measured at the community or ecosystem level
“/ Community functional parameter “/ Any feature resulting from the community-aggregation of functional traits

889
 . a trait is a feature measurable at the level of the Chapin, F. S. III et al. 1993. Evolution of suites of traits in
individual, which does not require additional response to environmental stress.
Am. Nat. 142: 78

information from the environment or at any other 92.
organizational level; Cornelissen, J. H. C. et al. 2003. Functional traits of woody
. a hierarchy is recognized among traits, expanding plants: correspondence of species rankings between field
Arnold’s (1983) ‘‘morphology, performance, fit- adults and laboratory-grown seedlings?
J. Veg. Sci. 14:
311
322.
ness’’ paradigm to plant ecology: performance
Darwin, C. 1859. On the origin of species.
John Murray.
traits are those which contribute directly to fitness, Dı́az, S. and Cabido, M. 2001. Vive la différence: plant
while functional traits are those morpho-physio- functional diversity matters to ecosystem processes.
phenological traits which have an impact on
Trends Ecol. Evol. 16: 646
655.
performance traits (and thus indirectly on fitness); Dı́az, S. et al. 2007. Functional diversity-at the crossroads
. integration functions among organization levels between ecosystem functioning and environmental filters.
should be made explicit when scaling-up to the
In: Canadell, J. G. et al. (eds), Terrestrial ecosystems in
levels of populations, communities and eco- a changing world. Springer-Verlag, in press.
systems. Dı́az, S. et al. 2006. Plant trait responses to grazing
a global
synthesis.
Global Change Biol. doi: 10.1111/
We believe that applying these definitions and the set j.1365.2686.2006.01288.a
of rules summarized in Table 3 will greatly enhance the Duarte, C. M. et al. 1995. Comparative functional plant
communication to address questions pertaining to the ecology: rationale and potentials.
Trends Ecol. Evol. 10:
impacts of global change drivers on species, commu- 418
421.
nities and ecosystems using trait-based approaches. Ellenberg, H. 1988. Vegetation ecology of central Europe, 4th
edn.
Cambridge Univ. Press.
Eviner, V. T. 2004. Plant traits that influence ecosystem
processes vary independently among species.
Ecology
Acknowledgements
The ideas in this paper come from 85: 2215
2229.
discussions that took place in the ECOPAR group of CEFE Eviner, V. T. and Chapin, F. S. III 2003. Functional matrix: a
during the preparation of the 2005 annual symposium of the conceptual framework for predicting multiple plant effects
Laboratory on ‘‘Life-history traits, functional traits and demo- on ecosystem processes.
Annu. Rev. Ecol. Syst. 34:
graphic traits in ecology and evolution’’ organized by P. David
455
485.
and E. Garnier. We thank the other speakers of this symposium,
Field, C. B. and Ehleringer, J. R. 1993. Scaling physiological
whose ideas helped us in the development of the paper.
processes: leaf to globe.
Academic Press.
S. Lavorel, C. Roumet and A. Kerkhoff provided useful
Garnier, E. et al. 2004. Plant functional markers capture
comments on various versions of the manuscript. This is a
publication from the GDR 2574 ‘‘Utiliterres’’ (CNRS, ecosystem properties during secondary succession.
France).
Ecology 85: 2630
2637.
Garnier, E. et al. 2001. Consistency of species ranking based
on functional leaf traits.
New Phytol. 152: 69
83.
Garnier, E. et al. 2006. Assessing the effects of land use change
on plant traits, communities and ecosystem functioning in
References grasslands: a standardized methodology and lessons from
an application to 11 European sites.
Ann. Bot. doi:
Ackerly, D. D. 2004. Functional traits of chaparral shrubs in
10.1093/aob/mcl215.
relation to seasonal water deficit and disturbance.
Ecol. Geber, M. A. and Griffen, L. R. 2003. Inheritance and natural
Monogr. 74: 25
44.
section on functional traits.
Int. J. Plant Sci. 164: 21

Ackerly, D. D. et al. 2000. The evolution of plant
43.
ecophysiological traits: recent advances and future direc-
Grime, J. P. 1974. Vegetation classification by reference to
tions.
Bioscience 50: 979
995.
Arnold, S. J. 1983. Morphology, performance and fitness. strategies.
Nature 250: 26
31.

Am. Zool. 23: 347
361. Grime, J. P. 1998. Benefits of plant diversity to ecosystems:
Calow, P. 1987. Towards a definition of functional ecology. immediate, filter and founder effects.
J. Ecol. 86: 902


Funct. Ecol. 1: 57
61. 910.
Caswell, H. 1989. Matrix population models: construction, Grime, J. P. et al. 1997. Integrated screening validates primary
analysis, and interpretation.
Sinauer Associates Inc. axes of specialisation in plants.
Oikos 79: 259
281.
Caswell, H. 1996. Second derivatives of population growth Hartwig, U.A. 1998. The regulation of symbiotic N2 fixation:
rate: calculation and applications.
Ecology 77: 870
a conceptual model of N feedback from the ecosystem to
879. the gene expression level.
Perspect. Plant Ecol. Evol.
Chapin, F. S. III 1993. Functional role of growth forms in Syst. 1: 92
120.
ecosystem and global processes.
In: Ehleringer, J. R. and Hodgson, J. G. et al. 1999. Allocating C-S-R plant functional
Field, C. B. (eds), Scaling physiological processes. Leaf to types: a soft approach to a hard problem.
Oikos 85:
globe. Academic Press Inc, pp. 287
312. 282
294.

890
Honek, A. 1993. Intraspecific variation in body size and Petchey, O. L. and Gaston, K. J. 2002. Functional diversity
fecundity in insects. A general relationship.
Oikos 66: (FD), species richness and community composition.
483
492.
Ecol. Lett. 5: 402
411.
Hooper, D. U. et al. 2005. Effects of biodiversity on Petchey, O. L. and Gaston, K. J. 2006. Functional diversity:
ecosystem functioning: a consensus of current knowledge. back to basics and looking forward.
Ecol. Lett. 9: 741


Ecol. Monogr. 75: 3
35. 758.
Huston, M. and Smith, T. 1987. Plant succession: life history Petchey, O. L. et al. 2004. How do different measures of
and competition.
Am. Nat. 130: 168
198. functional diversity perform?
Ecology 85: 847
857.
Irschick, D. J. 2003. Measuring performance in nature: Poorter, H. and Van der Werf, A. 1998. Is inherent variation
implications for studies of fitness within populations. in RGR determined by LAR at low irradiance and by

Integr. Comp. Biol. 43: 396
407. NAR at high irradiance? A review of herbaceous species.
Irschick, D. J. and Garland, J. 2001. Integrating function and
In: Lambers, H. et al. (eds), Inherent variation in plant
ecology in studies of adaptation: investigations of loco- growth. Physiological mechanisms and ecological con-
motor capacity as model system.
Annu. Rev. Ecol. Syst. sequences. Backhuys Publishers, pp. 309
336.
32: 367
396. Pywell, R. F. et al. 2003. Plant traits as predictors of
Jax, K. 2005. Function and ‘‘functioning’’ in ecology: what
performance in ecological restoration.
J. Appl. Ecol.
does it means?
Oikos 11: 641
648.
40: 65
77.
Jones, C. G. and Lawton, J. H. 1995. Linking species and
Quétier, F. et al. in press. Plant traits in a state and transition
cosystems.
Chapman and Hall Inc.
Keddy, P. A. 1992a. A pragmatic approach to functional framework as markers of ecosystem response to past and
ecology.
Funct. Ecol. 6: 621
626. present land-use in subalpine grasslands.
Ecol. Monogr.
Keddy, P. A. 1992b. Assembly and response rules: two goals Reich, P. B. et al. 2003. The evolution of plant functional
for predictive community ecology.
J. Veg. Sci. 3: 157
variation: traits, spectra, and strategies.
Int. J. Plant Sci.
164. 164: 143
164.
Kerkhoff, A. J. and Enquist, B. J. 2006. Ecosystem allometry: Roche, P. et al. 2004. Congruency analysis of species ranking
the scaling of nutrient stocks and primary productivity based on leaf traits: which traits are the more reliable?
across plant communities.
Ecol. Lett. 9: 419
427.
Plant Ecol. 174: 37
48.
Killingbeck, K. T. 1986. The terminology jungle revisited: Sæther, B.-E. and Bakke, Ø 2000. Avian life history variation
making a case for use of the term resorption.
Oikos 46: and contribution of demographic traits to the population
263
264. growth rate.
Ecology 81: 642
653.
Knevel, I. C. et al. 2005. The LEDA traitbase. Collecting and Saugier, B. et al. 2001. Estimations of global terrestrial
measuring standards of life-history traits of the northwest productivity: converging towards a single number?
In:
European Flora.
Univ. of Groningen. Roy, J. et al. (eds), Terrestrial global productivity.
Lavorel, S. and Garnier, E. 2002. Predicting changes in Academic Press Inc, pp. 543
557.
community composition and ecosystem functioning from Semenova, G. V. and van der Maarel., E. 2000. Plant
plant traits: revisiting the Holy Grail.
Funct. Ecol. 16: functional types
a strategic perspective.
J. Veg. Sci.
545
556. 11: 917
922.
Lavorel, S. et al. 1997. Plant functional classifications: from Shipley, B. 2006. Net assimilation rate, specific leaf area and
general groups to specific groups based on response to leaf mass ratio: which is most closely correlated with
disturbance.
Trends Ecol. Evol. 12: 474
478. relative growth rate? A meta-analysis.
Funct. Ecol. 20:
Le Galliard, J.-F. et al. 2004. Physical performance and 565
574.
darwinian fitness in lizards.
Nature 432: 502
505. Shipley, B. et al. 2006. From plant traits to plant commu-
Mason, N. W. H. et al. 2005. Functional richness, functional nities: a statistical mechanistic approach to biodiversity.
evenness and functional divergence: the primary compo-
Science 314: 812
814.
nents of functional diversity.
Oikos 111: 112
118. Vile, D. et al. 2006a. A structural equation model to integrate
McGill, B. J. et al. 2006. Rebuilding community ecology changes in functional strategies during old-field succes-
from functional traits.
Trends Ecol. Evol. 21: 178
184. sion.
Ecology 87: 504
517.
McIntyre, S. et al. 1999. Disturbance response in vegetation
Vile, D. et al. 2006b. Ecosystem productivity can be predicted
towards a global perspective on functional traits.
J. Veg.
from potential relative growth rate and species abundance.
Sci. 10: 621
630.

Ecol. Lett. 9: 1061
1067.
Naeem, S. and Wright, J. P. 2003. Disentangling biodiversity
Walker, B. et al. 1999. Plant attribute diversity, resilience
effects on ecosystem functioning: deriving solutions to a
seemingly insurmountable problem.
Ecol. Lett. 6: and ecosystem function: the nature and significance
567
579. of dominant and minor species.
Ecosystems 2: 95

Noble, I. R. and Gitay, H. 1996. A functional classification 113.
for predicting the dynamics of landscapes.
J. Veg. Sci. Weiher, E. et al. 1999. Challenging Theophrastus: a common
7: 329
336. core list of plant traits for functional ecology.
J. Veg.
Noble, I. R. and Slatyer, R. O. 1980. The use of vital Sci. 10: 609
620.
attributes to predict successional changes in plant com- Wikelski, M. and Romero, M. 2003. Body size, performance
munities subject to recurrent disturbances.
Vegetatio and fitness in Galapagos marine iguanas.
Integr. Comp.
43: 5
21. Biol. 43: 376
386.

891
Willby, N. J. et al. 2000. Attribute-based classification of nitrogen concentration: generalisations across growth
European hydrophytes and its relationship to habitat forms and growth irradiance.
Oecologia 127: 21
29.
utilization.
Freshwater Biol. 43: 43
74. Wright, I. J. et al. 2004. The worldwide leaf economics
Whitham, T. G. et al. 2003. Community and ecosystem spectrum.
Nature 428: 821
827.
genetics: a consequence of the extended phenotype. Woodward, F. I. and Cramer, W. 1996. Plant functional

Ecology 84: 559
573. types and climatic changes: introduction.
J. Veg. Sci. 7:
Wright, I. J. and Westoby, M. 2001. Understanding seedling 306
308.
growth relationships through specific leaf area and leaf

892