Plant Functional Types in relation
to Disturbance and Land Use:
Synthesis and Challenges
GCTE Task 2.2.1. Workshop
Valencia 9th – 12th
May 2001
Organising committee
-
Sandra Díaz, IMBIV – Univ. de
Córdoba, Argentina
-
Sandra Lavorel, CEFE, CNRS
Montpellier, France
-
Sue McIntyre, CSIRO,
Brisbane, Australia
-
Juli G. Pausas, CEAM,
Valencia, Spain (local host)
-
Dutch Global Change Programme (NWO)
-
GCTE
-
Secretaría de
Estado de Educación, Universidades, Investigación y Desarrollo (SEEUID), Plan
Nacional de Investigación Científica, Desarrollo e Innovación Tecnológica
(National plan of Reseach, Development and Tecnological Innovation 2000-2003),
Spain
-
Centre National
de la Recherche Scientifique (CNRS), France
Executive summary
This workshop, which brought
together 40 scientists from the Mediterranean Basin, northern Europe, Latin
America, North America, Australia and New Zealand (see list in Appendix 1), was
the closing workshop of GCTE Task 2.2.1. The task has run for four years and
has produced some interesting conceptual, methodological and research
developments in the field of plant responses to disturbances and land use.
These achievements resulted from the activities of three networks initiated at
the Montpellier workshop in 1998 to address grazing, fire and land abandonment.
Conceptual advances were made during four workshops (San Diego 1997,
Montpellier 1998, Stockholm 1999 and Brisbane 1999) and a Symposium in the
latest International Rangeland Congress (Townsville, Australia, 1999). A
significant output has been a Special Issue in the Journal of Vegetation
Science.
The general aims of the
workshop were:
-
to bring together members of
the task and newcomers to advance some specific areas of work
-
to synthesize the work done
in the last years
-
to define future lines of
international action within this section of GCTE.
These aims were addressed
through a series of 10 plenary talks, 23 posters, and 8 discussion groups (see
detailed programme in Appendix 2). Specifically, comparative syntheses of
individual studies that have investigated functional attributes or types were
presented by the grazing and fire networks. The identification of PFTs related
to disturbance response has potential management applications, and discussions
addressed how to translate this knowledge into operational tools for managers.
Both fundamental and applied work still require improved approaches to data
collection, archiving, and analyses, as our knowledge of PFTs is still in its
infancy. It was widely recognised that an important area for new development
lies in linking disturbance response with ecosystem effects. There is therefore
a need to identify overlaps between traits associated with response and traits
that produce significant ecosystem effects.
Plans for future developments of PFT
work in GCTE
Grazing Network: The main product
of the compilation and synthesis of literature performed in 1998/2001 will be
an article, to be submitted to an international journal. The Network will
continue to exist, moving into a new phase (see report of group discussion for
details), under the leadership of Graciela Rusch.
Fire Network: The main product
of the actiivies performed in 1998/2001 will be an article, to be submitted to
an international journal. The Network will continue to exist (see report of
group discussion for details), under the leadership of Juli Pausas.
-
Working group on databases. There will be a
first workshop to discuss general policy and protocols to gather databases and
/ or establish a network of databases. The workshop will be held in Europe in
late 2001 or early 2002. Prospective leaders: Hans Cornelissen, Eric Garnier,
Michael Kleyer.
-
Modelling workshops. The first
workshop will be organized in Australia by Ross Bradstock. The modelling
workshop series is proposed as a contribution to the future Modelling
Activities in the IA.
-
Working group on methodology
for data analysis. This will be coordinated by Valerio Pillar.
-
Developing a work plan for
the integration of response and effect classifications. This topic, which
was recognised as a high priority across working groups, needs to be developed
in collaboration with Focus 4 in particular. Acknowledging that part of the
relevant community was not present at this workshop we propose that 1) a plan
for a new network needs to be devised across GCTE, and 2) a specific workshop
will be held to achieve this and firm up leadership. Follow up on this,
including for prospective leaders, will be overseen by Sandra Díaz and Sandra
Lavorel.
-
Other topics of interest:
-
several participants expressed interest in the development of PFTs in
relation to desertification. This should be coordinated by the Desertification
Activity of GCTE (Jim Reynolds).
-
Cross-cutting topics of interest for future attention include:
regeneration traits and strategies (including resprouting) and plant-animal
interactions.
Considering the burgeoning
activity of the networks, and recognising the needs for continued conceptual
and methodological integration on the topic of PFTs we propose that a full
Activity be devoted to this topic as part of the future Land Programme of IGBP.
This Activity would consist of a series of networks organised in a 'concentric'
manner, as follows:
-
Core networks: 1) continued work in the case of grazing and fire, 2)
consolidation and full activity in the case of land abandonment, and 3)
initiation for response and effects;
-
Satellite networks with links to other parts of the IA and GCTE: data
bases and methodologies;
-
Operational networks within other Activities: modelling (modelling
Activity), desertification (Desertification Activity);
Sandra Díaz is considering leadership for the new Activity, with an
agreement on a one year transition (2002) overseen by Sandra Lavorel.
Summary of working group discussions
1.1. Grazing synthesis: what does
grazing do to plant traits?
Participants:
F. Casanoves, S. Cousins, S. Díaz (chair), R.
Erjnaes, V. Falczuk, Z. Henkin, R. Lindborg, S. McIntyre, Y. Ozem, B. Peco, G.
Rusch, C. Skarpe, M. Sternberg
Preliminary results of a meta-analysis of the
effects of grazing on plant traits at a global scale was presented in one of
the plenary talks (What does grazing do to plant traits? By Díaz, Falczuk &
Casanoves). This analysis included more
than 200 studies from 6 continents stratified over latitude, rainfall and types
of grazing history. The synthesis indicated that there are very few traits
consistently associated with grazing at the global scale (i. e. canopy height,
growth habit, and life history) but the limitations of the data available
prevent this being a firm conclusion. The results will be compiled in a
publication. One of the conclusions is that the information available in the
literature strongly limits our capacity to test general models of grazing at
the global scale. A major gap is the lack of a systematic assessment of traits
in different biomes. The group strongly supported the continuation of the
Grazing Network, with a future focus on additional and systematic trait data
collection. The new data gathering phase should focus on a core list of key
plant traits and should be guided by standardized field protocols. In order to
counteract the current regional biases in datasets, it is crucial to
systematically record traits (even when they are absent). The traits to be
considered in the new phase should be related to three issues that have not
been effectively addressed in the existing literature: Palatability, grazing
and ecosystem function, and grazing and regeneration.
A preliminary list of core traits was proposed but a continued review is necessary bearing in mind the grazing-related checklist published in McIntyre et al. (1999):
Specific Leaf Area (SLA)* Leaf
area
Leaf toughness* Canopy
height*
Evergreen vs. decidous* Position
of dormant buds
Vegetative reproduction Resprouting
capacity*
Seed mass* Seed
shape
Seed dispersal structures (no consensus) Spininess
Taxonomic information (e.g. species, family)
* All these traits were viewed as important,
but those with asterisk were considered the most crucial ones. When traits
cannot be measured for all species, the most abundant ones should be given
priority.
The meta-analysis also detected gaps on
environment data. Ancillary site information has to be reported with detail
enough to interpret plant trait distributions. The group recommended that the
following information be reported in trait studies: geographical coordinates
(latitude and longitude), climate according to Koeppen’s classification, a
coarse estimation of site productivity (e.g., from global biome
classifications, IBP reports, etc.), rainfall, grazing pressure, type of
grazer.
1.2. Fire synthesis
Juli Pausas
(chair), Ross Bradstock, Bill de Groot, Neal Enright, Bill Hoffman, Jill
Johnstone, Francisco Lloret, George Perry
The discussion flowed on the line of the trait
analysis in relation to fire presented in J. Pausas’ talk. Three subjects were
considered in the discussion:
(1)
Trait co-occurrence (relation of fire traits with
other plant traits; correlative approach). How relevant are the hypotheses on
trait co-occurrence for systems with low fire recurrence? A preliminary
analysis of the importance of these hypotheses for different ecosystems
(Australia, Mediterranean Basin, Savanna, and Boreal) was addressed on the base
of the experience of the group members.
(2)
Plant traits and fire regime (dynamic approach). Fire
acts as a community-level filter, and so, different traits should be found in
different fire regimes. To test this, a comparison between plant traits
selected under high and low fire recurrence was undertaken for different
ecosystems (as above), considering high and low as a relative value, in
relation to each ecosystem. The final table showed the relative importance of
different traits in the different ecosystems, and it is a nice set of
hypotheses to test against field data.
(3)
Feedbacks. Feedbacks between fire regime and species
traits are crucial (e.g., fuel accumulator, fire supressor/enhancer, changes in
combustibility) and has not been addressed properly. Many models of ecosystem
dynamic do not consider this feedbacks, which can change the direction or
intensity of dynamic trends. Both, positive and negative feedback occur.
We acknowledge that
tropical rainforests are missing from our data and should be addressed in the
future (some are newly exposed to fire).
It was concluded that it
would be desirable to explore the trait –fire regime relationship from
different ecosystems using a common modelling framework, and explore, if
possible, the +ve/-ve feedback effects.
1.3. The use of PFTs in management applications
Participants: Bruce Campbell, Pablo Cruz, Valerie Eviner, Sandrine Jauffret, Michael Kleyer, Frédérique Louault (chairt), Valerio de Patta Pillar
The group discussed how PFTs could be useful
for managers, and recognized the importance of adapting the approach to
different management scales, objectives and tools .
a) Pasture management
Grassland
systems can be characterized by a large diversity of plants which is mainly
determined by the availability of nutrients and the degree of defoliation. The
manager needs to manipulate the grassland by controlling nutrient supply and
defoliation regime.
PFTs can be defined in terms of their position
along fertility and defoliation gradients. For each PFT, agronomic
properties(ability to grow in summer, grazing tolerance, etc.) and dynamic
properties (ability to respond to a change in the regime of perturbation, etc.)
can be identified. Traits chosen need to be easy to recognized in the field for
them to be useful to managers.
b) Landscape management:
Simulations of landscape processes allows
testing scenarios, which can be used by decision makers, e.g. in environmental
impact assessments. The PFTs used in the model described by Kleyer et al.
(1999) were determined from database analyses combining species and
environmental variables (water table, soil characteristics, fertility, level of
disturbance, etc.). Simulations were then developed using maps describing real
environmental conditions and disturbance regimes, and the model then provided
the most probable distributions of the PFTs in the landscape. Results of the
simulation were linked to the accuracy of the dynamic properties available in
the database for each PFT.
How PFTs can be useful to managers ? First
observations
- At each
level, managers need predictions from simulation modes using PFTs and their
dynamic properties, but :
PFTs have to be adapted to the
needs of the manager
PFTs have to be recognized by
the manager
- At the largest scale, managers are mainly
using simulations outputs, not PFTs directly. PFTs are used by the modelers to
build models or expert systems. Nevertheless, we need better knowledge of the
functions and/or indicators used by the managers in order to include them in
the models and in the research programs.
Future actions:
More work
is needed on PFTs that are relevant to management at different scales:
- workshops
involving managers, to identify their questions and needs.
- which are their objectives ?
- which functions are of
importance ?
- which are the indicators and
the rules they use for taking decisions ?
- is there a common list of PFTs that are both
relevant to managers and of significance to ecologist?
This collection
of information would then allow the following to be more accurately selected:
the traits
the
functions and properties that need to be evaluated for each group
There is a need for better links between traits
and functions. Second, there is also a need for better links between PFTs and
functions. In both cases, functions refer to both ecosystem functions and
functions of importance for the manager. Third, a comparison framework is
needed in order to link various studies in different situations, including some
common indicators and variables describing environmental conditions.
1.4. Using
disturbance response traits to predict changes in ecosystem properties?
The working
question asked was:
What are
the traits useful to predict each of the properties and how do they (or not)
match with disturbance response traits?
It was
first noted that (i) the relative importance of different traits and properties
will depend on the type of ecosystem, (ii) the disturbance regime should be
characterized by its frequency, intensity, total extent and spatial
heterogeneity, and (iii) the disturbance traits relate to regeneration, survival
(avoidance) and vegetative regeneration (tolerance), likely to be specific for
different types of disturbance.
Three
ecosystem properties were examined, with the main following conclusions:
1.
Community
species richness
-
framework:
hump-backed relationship, providing a focus on disturbance response traits on
the one hand, and competition traits on the other hand
-
role
of context: productivity and evolutionary history
-
role
of disturbance characteristics: frequency, intensity, extent and spatial
heterogeneity
-
role
of the species pool
Species
richness is in fine related to
intrinsic richness within the functional types selected by disturbance regime
(species pool) rather than disturbance response per se.
2.
Aboveground
primary productivity
-
Is
relatively well related to disturbance regeneration traits
-
Also
depends on competition traits that determine community structure
3.
Resistance
and resilience
-
Are
negatively correlated properties and hence depend on mostly independent traits
-
Resilience:
ruderal traits
-
Resistance:
stress tolerance traits
-
Resistance
(growth related traits) is likely to be more disturbance specific than
resilience (regeneration traits mostly)
-
Importance
of feedbacks to disturbance regimes through changes in ecosystem properties
such as flammability, palatability.
We suggest
that a a joint workshop with Focus 4 is needed to develop a working plan to
address the question which now appears as a priority.
2.1. Convergence of classifications for grazing / fire / land use change
In order to
examine traits in common between responses to different disturbance types we distinguished
disturbance traits relate to:
-
Regeneration:
little specificity with respect to disturbance type ?
-
survival
(avoidance) and vegetative regeneration (tolerance): more specific to type of
disturbance?
-
-
Regeneration:
little specificity with respect to disturbance type ?
-
survival
(avoidance) and vegetative regeneration (tolerance): more specific to type of
disturbance?
As a first
study case, we compared fire and grazing. The approach taken was to examine
individually traits that are recognised as relevant to fire (from the list
provided by the fire discussion group) and those relevant to grazing (from
participants of the grazing working group). These traits were broken into three
broad functions that determine disturbance response: avoidance, tolerance and
regeneration.
|
Function |
Fire |
Grazing |
|
Avoidance |
High RGR Bark
thickness Tall
height Soil seed
bank, canopy seed bank? Phenology |
low SLA NA except
for large browsers Short
height (grazing), Tall height (browsing) Soil or canopy
seed bank Small
leaves Physical
defence: Leaf toughness, spines, hair… Phenology Chemical
defence Canopy
architecture Tissue quality |
|
Tolerance |
High RGR Resprouting
(above or belowground) |
High RGR Resprouting |
|
Regeneration |
(High
RGR) Canopy or
soil seed bank Light
demand (of seedlings) vegetative
reproduction fire
stimulation of flowering / germination dispersal vegetative
reproduction seed mass |
high SLA soil seed
bank (grazing), canopy seed bank (browsing) Light
demand (of seedlings) vegetative reproduction dispersal vegetative reproduction seed mass |
-
The
synthetic table indicates that our initial hypothesis was mostly correct.
Regeneration traits are largely shared across disturbance types, while traits
associated with avoidance can be more specific. Tolerance may be an
intermediate case.
-
Short
lists of traits will differ whether they are made to fit a single disturbance
or whether they are made to fit several disturbance types.
2.2. Data bases of plant functional traits
Participants: Sandra Díaz, Valerie Eviner, Eric Garnier, Michael Kleyer, Begoña Peco, Hans Cornelissen (Chair)
First we
listed for which exercises the global database setup can be useful:
(1) as a reference for the future, to
test yet unformulated questions
(2) to supply data or derived output to
endusers for particular purposes (eg land management, see Bruce)
(3) comparing functional plant traits
across biomes (fundamental tradeoffs)
(4) to identify consistent syndromes of
traits à
PFTs
(5) differences within (3) or (4) among
biomes, across environmental gradients, in response to global changes and/or
disturbances, etc.
(6) to test issues of scale (grid sizes;
local, regional, global)
(7) to compare responses and effects of
traits
(8) feeding functional traits or PFTs
into global models for carbon, nutrient, water budgets, etc. (see 2000 Avignon
workshop)
(9) linking to other databases (eg.
taxonomic, species effects, species distribution, productivity, decomposition)
Practical aspects of developing a global
database network:
It was not
considered feasible to aim for one worldwide database. A workable alternative
could be a flexible large network of (regional, national, supra-national)
databases.
However,
this could only work if database holders are happy about the arrangements with
respect to publications resulting from exercises using the database network. In
practise this comes down to authorship issues. The following setup was
suggested:
(a) Individual groups would maintain and
hold their own databases, but would agree to make these available for certain
uses within a GCTE network of database holders; and make these physically
available in the right format.
(b) Individuals or groups could use the
global database network to tackle a particular problem if they got permission
from GCTE; for this we proposed to create a steering committee whose members
would themselves mostly be important data providers.
(c) One possible mechanism (aimed at
keeping both journal editors and data suppliers happy) would be to follow
recent examples of GCTE-related meta-analysis papers, where there are two tiers
of authors. The first tier would contain (for instance) up to ten authors, of
which the last 'author' is the acronym for the network of data suppliers.
(d) A group of people should develop a
meta-database which contains the most important features of the databases
worldwide that can and agree to participate in this network.
(e) A workshop is needed to:
1.
form
the steering committee, with a GCTE stamp of approval;
2.
compile
all metadata for the databases identified for the network;
3.
build
the meta-database;
4.
set
the 'rules of the game' (authorship etc.);
5.
define
the exact format in which the GCTE-available version of each database should be
to make it ready for use within the database network;
6.
publicise
this meta-database within the GCTE community and beyond.
2.3. Methodologies in plant
functional analyses
The group
discussion on methodologies in PFTs analyses (J. Leps, S. McIntyre, R. Ejrnaes,
B. Nygaard, F. Casanoves, V. Pillar - chair) examined the different
alternatives available and concluded that the analytical method should deal
with the traits, the communities and the factors ('3 tables' approach). Finding
emergent groups based only on species by traits matrix is not enough, for the
groups should be validated with community and environmental data in order to be
taken as functional. Furthermore, the method should cope with different
combinations of traits (PFTs), not only with traits, acknowledging that traits
are not necessarily additive. It was also pointed out that the methods should
be applicable in the joint analysis of communities from different floristic
regions.
2.4. PFT applications in
modelling
Participants: Bruce Campbell (chair), Pablo
Cruz, Zalmen Henkin, Sara Cousins, Bettina Nygaard, Frédérique Louault
Approach: focussed workshops on specific
systems :
1 :
Grassland management model, Site specific grassland system
-
Participants:
local farmers, agronomist, GCTE science community
-
Focus
on local site
-
key
issues for manager, ecosystem processes
-
what
are key processes to model
-
output
-
compare
generalisable traits across diverse sites (using knowledge or network)
Approach:
-
rule-based
model development (spread sheet)
-
capture
expert knowledge
-
specify
relationships betwen traits, sites variables and sites and ecosystems
properties of interest
-
emphasis
on gaps in knowledge
-
what
are the elements tranposable to other sites/issues
-
traits
will be emergent (not preset)
Time frame:
by end 2002 in Australia
2-
Landscape scale management , landscape scale modelling
Topic :
landscape biodiversity conservation for planners
Objective:
Develop « bottom-up » approaches using traits
Participants:
landscape management planners, GCTE Scientists
-
scale
: site to region
-
End
users specifiy the outputs
-
scientists
generate inputs of « traits+types+driving variables »
-
regression
functions incorporated in GIS databases
-
assign
site variables, analyse gradients (Michael Kleyer’s methodology)
-
issues
abiotic explanatory variables (historical, ...)
-
how to
link detailed demographic knowledge for some species to PFTs ?
- &n