Biodiversity in a changing world : land use and other effects on landscape and
regional diversity
Sandra LAVORELLaboratoire d’Ecologie Alpine, CNRS
Grenoble, FRANCE
Lecture outline
• Global change effects on biodiversity– Some observations – biodiversity change under our eyes– Projecting climate change effects on plant distributions– Projecting land use change effects on biodiversity
• Plant functional approaches: concepts and methodologies• Projecting land use change effects on plant diversity and
ecosystems• Assessing consequences for ecosystem service delivery
A discernible « fingerprint » of climate change effectson species distributions and phenology
Type of change Changed as predicted Changed opposite to prediction
Phenological 87% 13%
Distributional changes
At poleward/upper range boundaries 81% 19%
At equatorial/lower range boundaries 75% 25%
Community (abundance) changes
Cold-adapted species 74% 26%
Warm-adapted species 91% 9%
Overall 81% 19%
Meta-analyses
Range-boundaries 6.1 km / m per decade northward/upward shift
Phenologies 2.3 days per decade advancement
Parmesan & Yoh 2003
Under our eyes: changes in biodiversity in response to climate change
Invasion by warm-climate plants
Num
bero
f spe
cies
19001980
2003
Species colonization on high summits
A first methodology: expert opinion
1 = land use, 2 = climate, 3 = N deposition,4 = invasions, 5 = CO2 (Sala et al. 2000)
Sala et al. 2000
Expert opinion points to land use as the public enemy
Population: the public enemy
Cincotta et al. 2000
High population density in biodiv hotspots and effects of increasing numbers of small households
Observed land use effects on biodiversity
Warren et al. Nature 2001
British butterfly speciesBlack circle: climate suitable, presentRed circle: climate suitable, absentBlue circle: climate unsuitable, present
Landscape fragmentation impedes climate response;species-specific responses
Historic
migration
Methodologies to project diversity responseto global change
• Niche based modelling• More statistical models: habitat change and species-area• Modelling species distributions based on plant traits
Methodologies to project diversity response to climate change:
Niche-based modelling
Occurrence points
Current range prediction
Future range prediction
Projection back into geographicalspace
Water availability
Deg
ree-
days
Model of niche in ecological dimension
Ecological niche modelling
Geographic space Ecological space
Niche-based modelling of species distributions
Current distribution Environmental variables
Models
Response curves
Realised nicheSimulated current
distribution
Climate change scenarios
Niche-based models
Potentialrealised niche
Futur potential habitats0 200 400 600 800
0.0
0.2
0.4
0.6
0.8
1.0
Winter precipitation
Pro
babi
lité
de p
rése
nce
Pro
babi
lity
of p
rese
nce
Winter precipitation
Growing Degree D
ays
Moi
stur
e In
dex
Simulated distribution
Castanea sativa
Futurs suitable
climatic space
• Data adequate (species distribution and environmental variables)
• Environmental variables control geographic range limits
• Statistical methods detect correct relationships
• Climate and LU will change as predicted…
• … and the same species-environment relationships will hold (no plasticity or genetic adaptation)
• Niche conservatism
Niche-based models
Primary assumptions
2,3 – 18,2Max CC
2,9 – 16,3Mean CC
2,0 – 15,1Min CC
Extinction rates
Extinction risks for:
1350 species, 8 models
3 CC scénarios
8 fold!!
- Compiling of 18 distinct analyses- Different spatial scales- Different niche-based models- Different environmental variables
- Different climate change scenarios
Production species and aesthetic value: Larch (Larix decidua)
‘Economic’ scenario 2080
‘Environmental’scenario 2080
Conserving Europe’s natural capitalCommitments for biodiversity under Natura 2000
Pulsatilla alpina Aquilegia alpina
Impact on plant species richness
2080 - A1 HadCM3
Future – present richness
Method: Aggregating individual species responses to projectchanges in species diversity
Relative sensitivity of plant biodiversityto climate change (2080)
Mountain and mediterranean regions are the most threatenedregions in Europe
In mountains greatestchanges are expected atintermediate altitudes (e.g. Prealps)
Modelling effects of land use on biodiversityThe Millennium Ecosystem Assessment:
habitat change and species-area
S = CA z
Habitat availability
Estimated species loss
Biome-specificspecies-area relationship
Sala et al. 2005 MA
Effects of land use change on biodiversityMillennium Ecosystem Assessment projections
Including effects of climate changeand N deposition
Sala et al. 2005 MA
Relative sensitivity of different biomes to different factors
Sala et al. 2005 MA
Bringing some biology into models:Modelling species distributions
based on plant traits
Plant functional traits
• A tool to reduce the complexity of life…• A functional rather than taxonomic perspective
– Groups of species or populations• with a similar RESPONSE to environmental change• and/or similar effects on ecosystem function…
– … as a result of shared characters– Functional traits:
• morphology, ecophysiology, demography, biochemistry
Diaz (2001)
Wright et al. Science 2004
Nitroge
n conte
nt
Leaf mass area
Phot
osyn
thet
icra
te
2550 species175 sites
‘The world wide spectrum of leaf economy’
Traits as markers of plant function
Density, diameterSpecific root length
Absorption (nutrients, water)Carbon fluxes (exsudation…)
Plant canopy heightOther?
Light interceptionCompetitive ability
Soft trait
Seed massSeed characteristics
Function
FecundityDispersalEstablishment
Resorption of nutrients;decomposability of litter
Traits of living leavesNIRS spectrum; other?
Traits for ‘easy’ measurement on large numbers of species in the field
Towards a short (?) list of soft traits
Cornelissen et al. 2003 Aus. J. Bot
Leaf area on an aridity gradient: Functional trade-off between stress tolerance and productivity for leafdesign
Predictive value for regionaldistributions and climate response
Thuiller et al. 2004 Ecology
Effects of traits on species distributionsDistribution along regional climatic gradients
88 Leucadendron species in a 1x1 km grid
Leaf area (log)
OM
I axi
s1
2.0 2.5 3.0 3.5 4.0 4.5
-2-1
01
23
Arid
itygr
adie
nt
Leaf area (log)
Species niche position on the aridity gradient
Trait responses to climate at the global scale
Wright et al. 2005 GEB
RadiationRai
nfal
l
Leaf
N
Consistent average relationship :
=> selective pressures associated with adaptation to different climatic regimes.
High local variability
Using plant traits to model species distributions and diversity at the global scale
Plant Functional Groups to projectvegetation distributions at global scales
Neilson et al. 2005
Simulating species diversity based on traits
Kleidon & Mooney GCB 2000
Plant traits to explain community distributions within landscapes:
Projecting land use change effects on plant diversity and ecosystems
Explaining community distributionsusing plant traits:Methodology
• Community responses to environmental gradients– Changes in species composition– and/or changes in species traits– determine changes in community-level ‘aggregated traits’
∑=
=n
iiiagg traitptrait
1
*
calculated for species whose cumulated biomass represents at least 80% of the community maximum standing biomass
Biomass ratio hypothesis (Grime 1998): species effects on ecosystemsare proportional to their relative abundance.
Relative contribution of species i to the maximum biomass of the community
Trait value of species i
Garnier et al. 2004
FP5 VISTA
Vulnerability of Ecosystem Services to Land Use Change in Traditional Agricultural Landscapes
11 sites in 9 countries…in marginal agricultural areassubjected to rapid change:
decreased management intensityabandonment
VISTA sites: a climate gradient across Europe
Aridity
Israel
Garnier et al . 2007 Ann. Bot.
Rai
nx
Tem
p.
Cold when it rains
Mild temp. + high rainfall
Variation in Leaf Dry Matter Content with climate
Mean annual site temperature (°C)
0 5 10 15 20 25
Ave
rag
e si
te L
DM
C (
mg
g-1
)
180
200
220
240
260
280
300
320
340
r = - 0.63*
Aggregated LDMC per site
Same correlations than at species level (Wright et al. 2005)
Colder sites hard more ‘dense’ leaves
=> prevalence of more annual species (with lower LDMC) at warmer (e.g. Mediterranean-climate) sites
LDMC: ratio dry to fresh mass of leaves => density in structural tissue
Response of community-level LDMCto decreasing land use
Wald statistic = 8.45, P < 0.001
Increase in LDMC in response to less intensive land use
Strongest at the warmest sites
Decreasing Mean Annual Temperature
Garnier et al . 2007 Ann. Bot.
Extensive use
Intensive use
Trait responses to decreasing land use:Generalizable across VISTA sites
Extensification promotesplants that conserve mineral
resources
Extensification promotes tall plants
Site Trait VPH RPH Flowers Seed M SLA LDMC LNC LCC LPC St DMC Clonality France mon ** m m m ** *** * NS * *** Scotland NS NS * ** * * NS NS NS Germany *** *** NS * * *** *** NS France Pyr * NS ** NS * ** Portugal *** m m *** *** NS *** m m Greece NS * ** NS ** m * NS *** * * Sweden * * *** NS NS ** NS NS NS ** * Norway ** ** ** ** Czech ** ** NS * * NS Israel ** NS NS NS NS France Alp *** ** NS * *** *** *** *** *** *** *** m = marginally sig.
Soil covariates: These trait responses to land use result from declining fertility
Garnier et al . 2007 Ann. Bot.
From plant traits to ecosystem functioning
PFT to scale from communities to ecosystems
?
Environmentalchanges
Responsetraits
Communitystructure and
diversity
Effecttraits
EcosystemfunctioningChallenge:
Linking responseand effects
Chapin et al. Nature 2000Lavorel & Garnier Funct. Ecol. 2002
Effects of plant traits on ecosystem properties:
(-) land use: P < 0.001 (-) LDMC: P < 0.001Land use x LDMC: P = 0.007Land use effects greatest at sites with highest LDMC
Garnier et al . 2007 Ann. Bot.
Effects of community LDMC on litter pools
Accu
mul
ated
litte
r
Litter decomposability in vitro:relationship with LDMC
r = -0.56***(n = 105)
LDMC aggregated (mg g-1)
0 100 200 300 400 500 600
Lit
ter
dec
ay r
ate
(g k
g-1
d-1
)
0
5
10
15
20
Sweden
Israel
Fortunel et al. In prep.
Without climate/land use/soil effects
=> litter with higher LDMC decomposes less well
VISTA: Putting the picture together
Decreasingland use
Increasing LDMC
Decreasingdecomposition
Litteraccumulation
Declining fertility
=> Chain of correlation (hopefully causality) and feedbacks
Soil C:N
Nitrogen NI
P Olsen
SLA
LNC
LPC
LDMC
P NI
Land use
Litter
Max Biomass
Decomposition
Linking land use effects on soils,plant traits and ecosystem properties
Lavorel et al. unpublished
Soil Phosphorous Content
Phosphorous available for plant
Leaf phosphorous content
Nitrogen available for plant
Modelling changes in landscapes and ecosystem properties in response to land use
change1920
2003
?2050
Modelling dynamics at the landscape scale
4 PFTs represented by 4 dominant grass species: Festuca paniculata, Sesleria caerulea, Bromus erectus & Dactylis glomerata
Landscape modelling plateform LAMOS
Parameters estimated through experiments and field observations
Validated using historical land use data:Calibration under past conditions and simulation/validation under current conditions
Fertility
UnploughedPloughed
Mowing
No mowing
Limited recruitment of Festucapaniculata
A1
A1
21A3
C1C2
C2 C2
3b1A3
C2 C2C2C2
C2B1A2A3
B1
B2
B2
C1 C1
C2
A1
A3
A3
A2A2
Festuca density
Fertility
Landscape dynamics modelling using LAMOSValidation against field data
LAMOS
F. Quétier et al. subm.
Projecting scenario impacts using LAMOS
Initial Sesleria PFT abundance
Initial Festuca PFT abundance
Sesleria PFT abundance under A1 scenario
Festuca PFT abundance under A1 scenario
Projecting effects on ecosystem services
Stakeholders
Ecosystemservices
Ecosystemdescriptors
Ecosystemproperties
Plant traits
Scientists
Landscape model
Ecosystem services identified by stakeholdersand proxies to quantify them
Quétier et al. submitted
Statistical relationships between functionalcomposition and ecosystem properties
Quétier et al. submitted
Scenario projections of ecosystem services
Quétier et al. submitted
Main take home messages
• Different modelling tools can be applied to project changes in biodiversity– Conceptual models: e.g. Plant Functional Traits– Statistical models: e.g. niche-based models, trait-ecosystem– Dynamic models: e.g. landscape-scale models
• These need to be based on solid data bases and experimental data
Strong links between modellers and field ecologists
Do not forget: this is not reality, models are tools to understand the present and analyze the future!
Geographic distribution of different types of extinction risk
93% species will have overlapping distributions
2% will not have overlappingdistributions
5% will lose their habitat entirely
Study site: Lautaret south facing slopes
A changing land use
A well known site
Few trees by now
Will land use change affect the dynamic and colonization of trees in the study area?
Cotoneasterjuranus
SorbusaucupariaLarix
decidua L.Rosa
glauca/pendulina
Sambucusracemosus
Prunus padus
Tree colonization in mountain landscape
Absence Increasing Probability of presence
South facing grasslands of Villar d’Arêne are potentially suitable for Larix decidua
Ecological niche-based modelling
Several parameters required for each PFTMost found in literature: life span, maturation time, potential seed productivity, ....
Larix produces seeds every 10 years
Germination rate at different light level was determined with experiments
5 light levels 3 light types (L, V, A) 2 water levels50 seeds per treatment
Some sensitive parameters stay Unknown: we tested them with simulations
Tree PFT: Larix decidua
litter shadow (L) / plant shadow(V) / artificial shadow(A)
germination rate in different light and water treatment
0
0,2
0,4
0,6
0,8
1
1,2
100 A36 V35 A26 A10 L7 V5 L1 V0,1 L0 A0
light and water treatment
ger
min
atio
n r
ate
water
no water
LIGHTGERMINATION
Tree dynamics at landscape scale
Simulations with Lamos
Dispersal capacity Plant ability for resource uptake Interaction of seedlings and juveniles with resident vegetation
Are expected to affect the tree line response to a changing environment
Use of a factorial design combining:
- 2 dispersal modes: Long distance dispersal vs No long distance dispersal
- 2 resource uptakes: Large vs narrow niche breadth for productivity
- 2 juvenile responses to light competition: shade tolerant vs shade intolerant
- 3 types of disturbance: Early mowing / late mowing / no mowing
on a complex productivity gradient
24 simulations
Tree dynamics at landscape scale
3 types of response:
- A complete colonization(100 trees by pixel ~ ha)
- A colonization by patches
- No real colonization (a few trees everywhere) others
Results
Few treesPatchesColonized
Tree dynamics at landscape scale
No real effect of land use disturbance
No real impact on grassland composition
• Long distance dispersal• Large niche breadth for productivity• Shade tolerant Juveniles
Shade tolerant juveniles
Narrow niche breadth patches on high productive soil
Large niche breadth limited by dispersal