class: fullbleed, math background-color: black         .absolute.w-100pct.pa-2.center.t-40pct.ba.bw-0.br-0.bg-white-80pct[ ## <b>Predictive modelling of plant communities</b> <br> Challenges for a dynamical approach ] --- layout: true .footer[ <!-- - @DrIsaBlg --> - <i class="fab fa-github"></i>iboulangeat <!-- - isabelle.boulangeat@irstea.fr --> - october 2019, MONACAL Colloque, Amiens <!-- -  --> ] <!-- --> --- class: # Prediction in community ecology .absolute.l-4.t-4[**Why?**] -- .absolute.l-10pct.t-5[climate change<br>habitat fragmentation<br>land-use changes<br>changes in disturbances<br>changes of human practices] -- .absolute.r-6.t-4[**What?**] -- .absolute.l-50pct.t-5[species composition<br>species diversity (taxonomic, functional, phylogenetic)<br>community-level attributes (annual biomass production, canopy height, ...)] --- class: # Coexistence theories #### What explain the presence of a species in a particular place? --  --  --  --  --  <!-- ========================================================================== --> --- class: title, smokescreen, no-footer background-image: url(abandon_static.png) # Static approaches and their limits --- class: # The SDM framework  --- class: # Explain with SDM, an exemple  -- .absolute.r-2.b-2[Boulangeat et al. 2012, EcoLet]  --  --- class: # Explain with SDM, dispersal indice  --- class: # Explain with SDM, species interactions indice  --- class: # Including species interactions 1. Biotic elements as explicative variables (dominant species, community attribute, ...) -- 2. Joint modelling (jSDM) : conditional presence -- Static approaches **infer** biotic interactions (and resulting community assembly) from species' co-occurrences -- <br><br>-> Hypothesis of equilibrium for model calibration --- class: col-2 # Integrating stacked-SDM and assembly rules  --  .absolute.l-4.b-2[D'Amen et al. 2015, JOB and GEB] -- <br>Historical and evolutionary constraints <br>-> potential species pool ("Dark diversity") -- <br>Abiotic constraints <br>-> ex. SDM -- <br>Macroecological constraints <br>-> species richness and/or trait space -- <br>Species interactions (assembly rules) <br>-> random selection, MaxEnt, ... <!-- ========================================================================== --> --- class: title, smokescreen, no-footer background-image: url(abandon.gif) # A dynamic perspective --- class: # Response time  -- .fixed.bg-white-80pct.t-50pct.l-50pct[->colonisation credit] .fixed.bg-white-80pct.t-80pct.l-50pct[->extinction debt] --- class: # Origin of lags -- #####Rapid <br> global changes   IGBP and resilience center --  --  --- class: # Aims of dynamic approaches <br> 1. **simulate** species interactions and community assembly 2. predict/analyse transient states <!-- ========================================================================== --> --- class: title, smokescreen, no-footer background-image: url(dandelion.jpg) # Models based on metapopulation dynamics --- class: # Metacommunity framework  --- class: fit-h1 # Models based on metapopulation dynamics | STATES| **presence**| **absence**| | ------| ---------| --------| | **presence**| 1-p(e) | p(e) | | **absence**| N.p(c) | 1-N.p(c) | .absolute.fr.r-10pct.t-20pct.ofc.w-5-12th[e=extinction ; c=colonisation ; N=regional prevalence] --  --- class: fit-h1 # Models based on metapopulation dynamics  .absolute.r-2.b-2[Talluto et al. 2017, Nature Ecology and Evolution] --- class: # Integrating biotic interactions 1. Biotic elements as explicative variables (dominant species, community attribute, ...) 2. Regroup species 3. Multi-species model --- class: # Integrating biotic interactions Community level: 2 ~species .absolute.r-2.b-2[Vissault et al., in revision]  --   --  --  --  -- --- class: # Predictions: a simulation approach .absolute.r-2.b-2[Vissault et al., in revision] <br>  --  --- class: # Integrating trophic interactions STM and Biomass population model  --- class: # Equilibrium shift Herbivores induce a **shifts** at biome transitions  --- class: # Trajectories  --- class: # Impact of herbivores on stability Herbivores **slow down** the return to equilibrium  --- class: fit-h1 # Impact of herbivores on transient dynamics Herbivores induce **more vegetation changes** in response to climate change and can have **opposite effects** depending on climate conditions  --- class: # Multidimentionality of stability Trophic interactions increases the **multidimentionality** of the resilience  <!-- ========================================================================== --> --- class: title, smokescreen, no-footer background-image: url(melezein_ubac_freaux2.JPG) # Dynamic vegetation models (DVM) --- class: # The interest of DVM .absolute.r-2.b-2[Snell et al. 2014, Ecography] 1. Multi-species! -- 2. Dynamic (simulation models) : transient states and lags -- 3. Process-based interactions : potential to predict non-analog communities -- 4. Multi-scale : individual to landscape processes -- 5. Spatially explicit -- <br><br> Processes of interest: reproduction, establishment, growth, mortality --- class: compact # Dynamic vegetation models  **Forest gap models (stand models)** <br>JABOWA (Botkin 1972), FORET (Shugart 1884), ZELIG (Smith 1988), SORTIE (Deutschmann 1997) <br>Aim : optimize **wood production** for harvest <br>Principle : indiv. based models based on **competition for light** -- <br><br> **Forest Landscape Models (FLM)** <br> LANDIS (He 1999), LANDCLIM (Schumacher 2004), TreeMig (Lischke 2006), LANDIS II (Scheller 2007)  <br>Aim : account for landscape processes (fire, seed dispersal) <br>Principle : upscaling stand models - Cohorts, height classes, PFT, representative cells, ... - Aggregate spatial and temporal scales <br> --- class: compact, img-left # Beyond forest focus: an hybrid exemple with FATE-HD FATE-HD among DVMs <br><br>  .absolute.r-2.b-2[Boulangeat et al. 2014, GCB] -- A landscape model dealing with **forest and non-forests** - Response to climate (via Habitat model) - Vegetation diversity (PFG) - Simpified population dynamics (competition for light, dispersal, demography) - Semi-quantitatif (easy to parameterize) - Disturbances (fire, grazing, mowing) --- class: # FATE-HD model .absolute.r-2.b-2[Boulangeat et al. 2014, GCB]  --  .absolute.r-4.t-3[Germination <br> Recruitment<br> Growth<br>Survival<br>Fecundity] --- class: # FATE-HD model .absolute.r-2.b-2[Boulangeat et al. 2014, GCB]  .absolute.r-4.t-3[Germination <br> Recruitment<br> Growth<br>Survival<br>Fecundity]  --  --  --- class: # FATE-HD model .absolute.r-2.b-2[Boulangeat et al. 2014, GCB]      --- class: # Is it enough to represent biodiversity?  --  --- class: # Is it enough to represent biodiversity? <br><br>  --- class: # FATE-HD vs SDM (at equilibrium) Refine PFGs distribution inside their habitat limits  --- class: # Scenarios  .absolute.r-2.b-2[Boulangeat et al. 2014, Ecography] --- class: # Scenarios  .absolute.l-4.t-30pct[Intensification]  <!-- .fixed.b-2.l-4[<video height="400" autoplay><source src="intens.mp4" type="video/mp4"></video>] --> .absolute.r-5.t-30pct[Abandonment]  <!-- .fixed.b-2.r-4[<video height="400" autoplay><source src="abandon.mp4" type="video/mp4"></video>] --> .absolute.r-2.b-2[Boulangeat et al. 2014, Ecography] --- class: # Changes in regional diversity  .absolute.r-2.b-2[Boulangeat et al. 2014, Ecography] --- class: # Change in local diversity  .absolute.r-2.b-2[Boulangeat et al. 2014, Ecography] --- class: fit-h1, no-footer # PFG turnover or dominance changes?  .absolute.r-2.b-2[Boulangeat et al. 2014, Ecography] --- class: title, smokescreen, no-footer background-image: url(polarbear.png) # Challenges and perspectives for multispecies modelling --- class: # Modelling approach  --- class: fit-h1 # Issues #1 : How to represent the whole communities? Many modelling entities -> numerous parameters to fit (ex. covariance matrix), or to document for simulation models -- - **organisational levels** simplification: <br>(individuals->cohorts->populations->PFG->PFT->community attributes) -- - using **networks** to reduce the dimension of interactions (instead of grouping elements directly) -- - **focal species and community**: reduce number of interactions to N (number of species) --- class: fit-h1 # Issue #1 : How to represent the whole communities? Further problems <br>-> Different nature of interactions (trophic, facilitation, dispersal, habitat) <br>-> Time dimension on interactions (ontology, phenology) --- class: fit-h1 # Issue #2 : How to integrate processes? A certain number modelling entities -> still too many parameters to fit or to document for simulation models .absolute.r-2.b-30pct[_Biodiversity vs explicit processes,<br> a necessary trade-off_]  .absolute.r-2.b-2[Gallien et al. 2010, DID] --- class: fit-h1 # Issue #2 : How to integrate processes? A certain number of modelling entities -> still too many parameters to fit or to document for simulation models - **hybrid models** : implicit and explicit processes see Gallien et al. 2010, DID -- - **Multi-scale data integration** see Clark et al. 2009, Ecological Monograph ; Talluto et al. 2016, GEB --- class: no-footer background-image: url(marais_acide_lautaret.JPG) # Discussion