Modelling biodiversity-ecosystem functioning relationships in grasslands

HAL Id: hal-02812257

https://hal.inrae.fr/hal-02812257

Submitted on 6 Jun 2020

HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

Modelling biodiversity-ecosystem functioning relationships in grasslands

Jean-François Soussana

To cite this version:

Jean-François Soussana. Modelling biodiversity-ecosystem functioning relationships in grasslands.

Impact of community composition on ecosystem function in a changing environment. Centre for Population Biology, Centre for Population Biology (CPB). GBR., Jun 2009, Silwood Park, United Kingdom. 46 p. �hal-02812257�

Modeling Biodiversity-Ecosystem Functioning

Relationships in Grasslands

Jean-Francois Soussana soussana@clermont.inra.fr Grassland Ecosystem Research

INRA, Clermont-Ferrand, France

Outline

Part 1. Statistical approach

Response and effect framework Mesocosm C cycling experiment

A mathematical test of the framework

Part 2. Simulation approach

An individual centred model for linking biodiversity and ecosystem functioning

Model evaluation with grass monocultures Simulating BD-EF and dominance

Role of plasticity

Trait space analysis

1. Statistical approach

Global change

Land use change Global Warming Species invasion

Ecosystem functioning

Individual response Species dynamics in interactions Suding et al. 2008 (GCB)

Trait based Response&Effect framework

Environmental drivers

Dominant trait values

Species &

functional diversity

Complementarity Mass ratio

Primary productivity Ecosystem services (e.g. C sequestration) Trait convergence

Trait d

iverge nce

Biodiversity experiments Field

experiments

(after Lavorel & Garnier, 2002,

Suding et al., 2008; Diaz et al., 2007)

Response and effect framework

• At time t ₁

(after Suding et al., GCB, 2008)

Response and effect framework

• At time t ₂

(after Suding et al., GCB, 2008)

A mathematical test of the framework by Suding et al. (2008)

(Klumpp & Soussana, GCB, 2009)

Grassland mesocosms

exposed to

C labelled CO

.

Treatments

Pre-experimental low grazing

Pre-experimental high grazing

LL HH

low disturbance high disturbance

Field Mesocosm

Start of experiment

13C labelling LL HL LH HH

5 cuts yr^-1 No cut

(Klumpp et al., 2007 AGEE, Klumpp et al., 2007 Biogeosciences)

Disturbance induced changes in C cycling

0 600 800 1000 1200 R= -0.51 P< 0.001 B

(Klumpp, Falcimage & Soussana, 2007, AGEE; Klumpp, Soussana & Falcimagne, 2007, Biogeosciences)

Above-ground net primary productivity (g C m^-2 yr^-1) SoilC sequestration (g C m-2 yr-1 )

Cutting disturbance

COCO₂₂scrubberscrubber Compressor

Compressor

Steady state ¹³CO₂ labelling Grassland mesocosm experiment

i) Cutting reduces soil C sequestration

ii) Cutting reduces mean residence time of C in soil fractions >200 µ

MRT= 22 month

MRT = 31 month

Cutting disturbance

Disturbance induced changes in

aggregated leaf and root (rhizome) traits

SLA (cm2 g-1 DM)

100 150 200 250 300 350

LDMC (g g-1 DM )

0.10 0.15 0.20 0.25 0.30 0.35

Year

2003-2004 2004-2005 LNC (g N g-1 DM )

1.0 1.5 2.0 2.5 3.0 3.5 4.0

SL(m g-1 DM)

0 20 40 60 80 100 120

Dens (g DM cm-3 )

0.1 0.2 0.3 0.4 0.5

Year

2003-2004 2004-2005 RNC (gN g-1 DM)

0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0

A) Leaf traits B) Root and rhizome traits

Constant low disturbance (LL, z) Shift to high disturbance (LH, {) Constant high disturbance (HH, T) Shift to low disturbance (HL, V).

(after Klumpp & Soussana, GCB, 2009)

C sequestration (gC m-2 )

Separating direct role of disturbance from Response&Effect

(Klumpp & Soussana, GCB, 2009)

b₂, direct disturbance effect

aE₂, trait mediated effect

A role of functional diversity

Rao’s index of functional divergence (FD) (Mason et al., 2005, Leps et al., 2006)

Conclusions (statistical approach)

• Response&effect susccessfully tested

• A simple equation encapsulates:

– Direct effects of drivers – Response&Effect of:

• Aggregated traits

• Functional divergence of traits

• Drawbacks

– Often models with different traits are difficult to discriminate statistically,

– More studies needed to generalise

DI S CO

VE R DI S

CO VE

DISCOVER project (2006-2008)

Plant Plant diversity diversity

SoilSoil microbial microbial diversity diversity

Herbivory Herbivory Soil fauna

Soil fauna diversity diversity

2. Mechanistic simulation approach

Grassland experiments in DISCOVER project

Community

Community structure structure -

- plants, plants, soilsoil biotabiota

-- diversitydiversity (genetic(genetic, species, species, functional, functional))

Ecosystem functioning C, N, water cycles

ORE -ORE - Grassland, Grassland, BiodiversityBiodiversity andand Biogeochemical

Biogeochemical CyclesCycles

Land use/management manipulation for the same initial biodiversity

Jena

Jena GrasslandGrassland DiversityDiversity ExperimentExperiment Plant diversity manipulation

for the same management

Structure Inter-yr comp.

Dominance plots Succession plots

SLA, RGR,

Fecundity, allometry, allocation, # and lifespan of leaves, phyllocron,

Data Plant

Parameter

Coupled Model

Biogeochemistry

Prediction

• Density

•Composition

• Evenness

• Succession

Observation

• Carbon

•Nitrogen

•Water

NPP, NEP, Rh, P, ET, SD, Nmin…

Confrontation

Demogpraphy

C/N, leaf protein, Vmax

storage,retranslocati on nutrient uptake rates, litter quality,

Trait space analysis

Climate Soil texture

Initial C, N, P stocks in soil

Initialisation, driver

pla nts

soil

GEMINI

Parameterisation Development, Improvement Evaluation and sensitivity analysis

Modelling framework

Motivation

• How does the current structure and functional diversity translate into ecosystem processes?

• Is functional group diversity enough to explain diversity effects. How much species-specific information do we need?

• Can we distinguish between complementarity and selection effect in the model and assign processes to these effect groups?

• Do correlations between traits (trade-offs) force different species into

complementary resource use?

Grassland Ecosystem Model with Individual centred Interactions, (GEMINI)

Graphical front-end showing the tree of modules.

Simulated architecture of tillers.

F. arundinacea L. perenne

(Soussana et al., 2000)

At leaf level

Farquhar’s photosynthetic model Stomatal regulation

Coordination of leaf N content Leaf N distribution vs. light Acclimation to temperature, CO₂ Morphogenesis

Architecture

Plasticity for some traits

…

At root level

NO₃^-, NH₄⁺uptake vs.

acquisition in function

of N in plant/ soil concentration Morphogenesis

Architecture

Acclimatation to Temperature

…

At soil level

Inorganic N balance

SOM dynamics (four SOM pools)

Microbial turnover (two decomposer types)

At plant axis level

Assimilate partitioning

Functional balance between roots & shoots Reserve dynamics

Substrate dynamics

Root, shoot structure & shoot proteins dynamics

…

At population level

Self thining

Axis density dynamics

Density dependent recruitment Mortality

…

At community level

Radiative balance (Kubelka-Munk equations) Inorganic N balance (diffusion driven competition) Community dynamics

At ecosystem level

Primary productivity C and N cycles

GEMINI model

A_L = b₁ll^b2 W_L = b₃ll^b4 Max leaf length ll^max

θ

Internode radius ri*

Internode density di*

Max. and min. length li_max, li_min Activation:

shoot buds flower buds

Leaf angle

Internode χ

angle

Area allometry Weight allometry

Generalising shoot architecture

Jena biodiversity experiment (coll. Christian Wirth, MPI BGC)

(Under development)

Model parametrisation from trait measurements in grass monocultures

(with no major limiting factors)

54 parameters were determined for 14 grass species

(data basis, see Pontes et al., 2007 Func. Ecol., 2008, GFS; Maire et al., Func. Ecol. 2009)

Common garden study, Theix (288 micro-plots) Native grass species (2002-2006)

Block I Block II

Block III

– 13 perennial pasture grass species

• Major grasses in French Massif Central

• Seeds collected in regional grasslands

– Species diversity

• Monocultures of all species

• 2 and 6 species mixtures

– Disturbance and nutrients factors

• Two cutting frequencies

• Two N fertiliser supplies

Photo: A. Peeters Alopecurus pratensis

Anthoxanthum odoratum

Photo: A. Peeters

Arrhenatherum elatius

Photo: A. Peeters

Dactylis glomerata

Festuca rubra

Festuca arundinacea

Simulation of aboveground biomass production in monocultures by GEMINI

Comparison with common garden data at Theix

N supply: N-, N+

Cutting frequency: C-,C+

(Soussana et al., in prep.)

Simulation of dominance in 6 sp. grass mixtures

N supply: N-, N+

Cutting frequency: C-,C+

(Maire et al., PhD 2009) Comparison with common

garden data at Theix

Simulation of biodiversity effect in grass mixtures vs. monocultures

(Maire et al., PhD 2009) Simulations show transgressive overyielding

Emerging property: self-thinning

(Soussana et al. in prep)

• Probably a key mechanism shaping BD- EF relationships

• Largely ignored so far

• A priori: very difficult to control

experimentally,requires modelling

Role of plasticity?

Plastic state variable Type Processes Controlled by

Shoot:root ratio M Nutrient versus light acquisition Light and N limitation Nitrogen uptake rate P Down-regulation of uptake Internal state of labile N Root elongation M Exploration of new soil layers Degree of ZOI overlap Area-based leaf N content P Down-regulation of

photosynthesis

Shade tolerance via N effect on respiration

Affects N-demand and exerts feedback to root:shoot allocation

Incident light at a given height in the canopy light gradient

Shoot internode length M Shade avoidance Light

Leaf length M Shade avoidance Light

Specific leaf area M Shade tolerance Light

Branching intensity M Regulation of sink strength Light threshold

Tillering M Vegetative regeneration Internal C and N

substrate availability, light

• Plastic behaviour (baseline) through inbuilt plasticity rules based on optimality assumptions

• Static behaviour by suppressing plasticity rules (state variable becomes parameter)

Plastic behaviour in GEMINI

Modelling the role of plant functional traits and of their plasticity

Leaf N

= f(Light)

N uptake capacity = f(internal N) Tillering = f(internal C,N; light)

Internode length Ramification

= f(Light) Leaf morphology

(length)

= f(Light)

Root:shoot partitioning

f(light and N)

Suppressing tiller dynamics

Suppressing shoot and root morphogenesis

Suppressing coordination of growth

(Soussana et al. in prep)

Do we need to simulate plasticity?

A test by comparing model and degraded versions

Plasticity affects model performance

Reduced plasticity

(Soussana et al. in prep)

Trait space analysis

• Given the role of plasticity, are traits close to optimal?

• Does the model generate trade-offs among traits?

• Are these trade-offs different among

species?

Role of traits for maximizing fitness

Leaf Life Span

SLA

Leaf Life Span

SLA

Traits co-vary along plant specialization axes (Wright et al. 2004; Ackerly 2004)

Different traits form a syndrom

Traits cannot be tested individually by experiments because of their covariation The model allows these tests

?

(cross sensivity analysis)

Full factorial design: 240, 000 simulations

2600 2700 2800 2900 3000 3100 3200 3300 3400

30 35

40 45

50 55

60 65

100 120 140 160 180

Tiller Density

Plant height Leaf L

ife span

Analysing systematically plant growth in a 4-D trait space

Comparison of monocultures from 12 species at two N levels

Traits

Two leaf traits: leaf dry-matter content

Two stature traits: tiller density

-0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6

100

120 140160180200 0.16

0.17 0.190.18 0.210.20

Biomass (g/plant)

Phylochron (°day) LDMC

(g/g)

A single trait combination maximizes plant biomass

(Gross et al. in prep)

Example of LDMC vs. PH analysis for Arrhenatherum elatius

Is plasticity optimal?

Model shows that trait plasticity maximizes plant biomass (fitness)

(Gross et al. in prep)

(Gross et al. in prep)

Optimal trait values and their trade-offs

Ridges between traits in A. elatius

(Gross et al. in prep)

Slope of the ridge is a between trait trade-off

Trait space

A. elatius, N+

Analysing trait space across species

(Gross et al. in prep)

Some of the processes controlling simulated biomass

-Coordination between C and N capture

(Osone & Tateno 2008; Maire et al. 2009)

- Coordination of leaf photosynthesis

(Maire, Soussana et al. submitted)

-Minimizing allometric constraints

(Enquist et al. 2007, Savage et al. 2007)

...

(Gross et al. in prep)

Effect of trait variation on simulated

C:N ratio of substrates

(N-)

Observed SLA 350 ppm (cm²/g)

50 100 150 200 250 300 350

SLA max predicted (cm²/g)

50 100 150 200 250 300 350

350 ppm: r ² = 0.95 ***

700 ppm: r² = 0.75 ***

(N+)

Observed SLA 350 ppm (cm²/g)

50 100 150 200 250 300 350

SLA max predicted (cm²/g)

50 100 150 200 250 300 350

350 ppm: r ² = 0.94 ***

700 ppm: r² = 0.58 **

Trait response to elevated CO

Æ Optimal SLA declines under elevated CO

(700 ppm)

Æ Larger effect at high N supply

Conclusions (simulation approach)

• Successful model for predicting variation in primary productivity across grass species and in response to N and disturbance;

• Dominance and biodiversity effect are also simulated in sown grass mixtures;

• Model suggests major role of trait plasticity for fitness

– Trait values in mixtures?

• Perspectives

– Trait data bases will help supplying model parameter values – Full version with architecture and flowering soon ready

– Water stress effects (under development)

– Evaluation with Jena experiment data (ongoing)

– First tests showing model predicts trait responses to elevated CO₂

Acknowledgements

• INRA UREP (Clermont, France)

– Nicolas Gross (Post-Doc) – Vincent Maire (PhD student) – Raphael Martin (Engineer) – Bruno Bachelet (Engineer)

• UMR LIMOS (Clermont, France)

– David Hill (Prof.)

• MPI BGC (Jena, Germany)

– Christian Wirth

– Tanja Reinhold

– Hans Dähring