Do biotic interactions impact geographic range limits of species?

name: title
    layout: true
    class: center, middle, style1
    ---

# THE TITLE IS OBVIOUSLY MISSING

### [Kevin Cazelles](http://kevincazelles.fr), [Universty of Guelph](https://www.uoguelph.ca/)    [![:fontawe](github)](https://github.com/KevCaz/seminarInteractionsDistributions)

### Département de biologie, Université de Sherbrooke

#### October, 12th 2018

---
 name: style1
 layout: true
 class: style1
 ---
 # ABOUT ME
 <hr>

## Theory

- biogeography
      - co-occurrence
      - food webs
      - population dynamics
      - homogenezation

## Statistics, algorithms, data, software

.right[<img src="img/logo/inSilecoLogo.png" height="200" align="right" style="padding:12px"/>]

### .left[[inSileco: codes, ideas and methods in ecology](https://insileco.github.io/)]

---
    name: title
    layout: true
    class: center, middle, style1
    ---

# DO BIOTIC INTERACTIONS IMPACT GEOGRAPHIC RANGE LIMITS OF SPECIES?

### [Kevin Cazelles](http://kevincazelles.fr), [Universty of Guelph](https://www.uoguelph.ca/)    [![:fontawe](github)](https://github.com/KevCaz/seminarInteractionsDistributions)

### Département de biologie, Université de Sherbrooke

#### October, 12th 2018

---
    name: style2
    layout: true
    class: center middle, style2
    ---

# INTRODUCTION
 <hr/>
 ### Biogeography and biotic interactions

---
 name: style1
 layout: true
 class: style1
 ---
 # Biogeography: distributions
 <hr/>

.center[![:scale 100%](fig/biogeo/ecoreg.png)]

---
 # Biogeography: distributions
 <hr/>

.center[![:scale 100%](fig/biogeo/figmapw3.png)]

---
 # Biogeography: processes
 <hr/>

.center[![:scale 100%](fig/mech/figall1.png)]

---
 # Individuals
 <hr/>

.center[![:scale 100%](fig/mech/figall2.png)]

---
 # Individuals to population
 <hr/>

.center[![:scale 100%](fig/mech/figall3.png)]

???
    Interaction at this scale that one can actually envision interactions

---
 # Population to metapopulation
 <hr/>

.center[![:scale 100%](fig/mech/figall4.png)]

???
    Interaction at this scale that one can actually envision interactions

---
 # Populations to metapopulations
 <hr/>

.center[![:scale 100%](fig/mech/figall5.png)]

???
    Interaction at this scale that one can actually envision interactions

---
 # Metapopulations to metacommunities
 <hr/>

.center[![:scale 100%](fig/mech/figall6.png)]

???
    Interaction at this scale that one can actually envision interactions

---
 # Metacommunities through time
 <hr/>

.center[![:scale 100%](fig/mech/figall7.png)]

???
    We scaled up in space
    Let's sclaed up in time
    Lots of process to take into account !
    That explain why species are here... Lot's of processes le'ts recap

---
 # Metacommunities through time
 <hr/>

.center[![:scale 100%](fig/mech/figall8.png)]

---
 # Metacommunities through time
 <hr/>

.center[![:scale 100%](fig/mech/figall9.png)]

---
 # Biogeography: processes
 <hr/> 
 ## 1. Abiotic variables
 --

## 2. Dispersion
    --

## 3. Ecological interactions
    --

## 4. Historical factors
    --

# .center[[INTERDEPENDENT]()]

---
 # Biogeography: questions
 <hr/>

--
    ## How processes generate distributions?
    ## How to infer processes from distributions?

---
 # Biogeography: challenge
 <hr/>

.center[![:scale 100%](./fig/figdem0.png)]

---
 # Biogeography: challenge
 <hr/>

.center[![:scale 100%](./fig/figdem1.png)]

---
 # Biogeography: challenge
 <hr/>

.center[![:scale 100%](./fig/figdem2.png)]

---
 # Species Distribution Models (SDMs)
 <hr/>

.center[![:scale 100%](./fig/figgrad1.png)]

---
 # Species Distribution Models (SDMs)
 <hr/>

.center[![:scale 100%](./fig/figgrad2.png)]

---
 # Predicting species distributions
 <hr/>

.center[![:scale 100%](./fig/figdem3.png)]

---
 # Predicting species distributions
 <hr/>

.center[![:scale 100%](./fig/figdem4.png)]

---
 # Species Distribution Models (SDMs)
 <hr/>

.center[![:scale 100%](./fig/figgrad3.png)]

---
 # Species Distribution Models (SDMs)
 <hr/>

.center[![:scale 100%](./fig/figgrad4.png)]

---
 # Predicting species distributions
 <hr/>

.center[![:scale 100%](./fig/figdem5.png)]

---
 # Predicting species distributions
 <hr/>

## Species are assumed independent
    --

### ![:fontawe](file-text) Davis *et al*, 1998, *Nature*
    --

## Joint Species Distribution Models (JSDMs)

### ![:fontawe](file-text) Clark *et al*, 2014, *Ecological Applications*

### ![:fontawe](file-text) Pollock *et al*, 2014, *Methods in Ecology and Evolution*

### ![:fontawe](file-text) Zurell *et al*, 2018, *Ecography*

---
 # Predicting species distributions
 <hr/>

--
    ## Theory

### ![:fontawe](file-text) Holt, *Toward a Trophic Island Biogeography*, 2009. [![:fontawe](external-link)](https://press.princeton.edu/titles/9096.html)

### ![:fontawe](file-text) Gravel *et al.*, *Ecology Letters*, 2011.[![:fontawe](external-link)]((http://onlinelibrary.wiley.com/doi/10.1111/j.1461-0248.2011.01667.x/abstract)

###  ![:fontawe](file-text) Massol *et al.*, *Advances in Ecological Research*, 2017.[![:fontawe](external-link)]((https://doi.org/10.1016/bs.aecr.2016.10.004)

---
 # Biogeography: challenge
 <hr/>

.center[![:scale 100%](./fig/figdem6.png)]

---
 # Biogeography: challenge
 <hr/>

.center[![:scale 100%](./fig/figdem7.png)]

---
 # Questions
 <hr/>

### How interactions influence species distributions?
    ### How to infer the role of biotic interactions from co-occurrence data?

--
    1. How to integrate biotic interactions into SDMs?

--
    2. How the properties of ecological networks influence co-occurrence?

--
    3. Can we infer interactions from co-occurrence data?

???
    des why / des hypothèses et des how

---
 name: style2
 layout: true
 class: center, middle, style2
 ---
 # THEORY OF ISLAND BIOGEOGRAPHY AND BIOTIC INTERACTIONS
 <hr />
 ### How to integrate biotic interactions into SDMs?

---
    name: style1
    layout: true
    class: style1

---
 # MacArthur et Wilson (1963, 1967)
 <hr/>

---
 # MacArthur et Wilson (1963, 1967)
 <hr/> 
 .center[![:scale 100%](fig/mcaWil/figMW2.png)]

---
 # MacArthur et Wilson (1963, 1967)
 <hr/> 
 .center[![:scale 100%](fig/mcaWil/figMW3.png)]

---
 # MacArthur et Wilson (1963, 1967)
 <hr/> 
 .center[![:scale 100%](fig/mcaWil/figMW4.png)]

---
 # MacArthur et Wilson (1963, 1967)
 <hr/> 
 .center[![:scale 100%](fig/mcaWil/figMW5.png)]

---
 # MacArthur et Wilson (1963, 1967)
 <hr/> 
 .center[![:scale 100%](fig/mcaWil/figMW6.png)]

---
 # MacArthur et Wilson (1963, 1967)
 <hr/> 
 .center[![:scale 100%](fig/mcaWil/figMW7.png)]

---
 # MacArthur et Wilson (1963, 1967)
 <hr/> 
 .center[![:scale 100%](fig/mcaWil/figMW8.png)]

---
 # Model properties
 <hr/>

### 1. Colonisation/extinction dynamics
    --

### 2. Equilibrium
 --

### 3. Elegant, easy to build on it
 --

### 4. Biotic and abiotic constraints are missing

???
    ça marche bien... et très utilisé

---
 # Releasing the assumption of independence
 <hr/>

--
    ## Gravel *et al*, 2011, *Ecology Letters* [![:fontawe](external-link)](http://onlinelibrary.wiley.com/doi/10.1111/j.1461-0248.2011.01667.x/abstract)

--
 ### 1. Island without prey = no predator colonisation
 ### 2. Predator without prey = extinction

--
 ## Colonisation and extinction depend on the local community

???
    première étape pour lever l'hypothèse.

---
 # What did I do?
 <hr/>

.center[![:scale 100%](fig/gtib/figGTIB1.png)]

---
 # What did I do?
 <hr/>

.center[![:scale 100%](fig/gtib/figGTIB2.png)]

---
 # What did I do?
 <hr/>

.center[![:scale 100%](fig/gtib/figGTIB3.png)]

---
 # What did I do?
 <hr/>

.center[![:scale 100%](fig/gtib/figGTIB4.png)]

---
 # What did I do?
 <hr/>

.center[![:scale 100%](fig/gtib/figGTIB5.png)]

---
 # What did I do?
 <hr/>

.center[![:scale 100%](fig/gtib/figGTIB6.png)]

---
 # What did I do?
 <hr/>

.center[![:scale 100%](fig/gtib/figGTIB7.png)]

---
 # What did I do?
 <hr/>

.center[![:scale 100%](fig/gtib/figGTIB8.png)]

---
 # Results: the classical theory revisited
 <hr/>

.center[![:scale 100%](./fig/figgtib0.png)]

---
 # Results: the classical theory revisited
 <hr/>

.center[![:scale 100%](./fig/figgtib1.png)]

---
 # Results: the classical theory revisited
 <hr/>

.center[![:scale 100%](./fig/figgtib2.png)]

---
 # Results: the classical theory revisited
 <hr/>

.center[![:scale 100%](./fig/figgtib3.png)]

---
 # Results: the classical theory revisited
 <hr/>

.center[![:scale 100%](./fig/figgtib4.png)]

---
 # Conclusion and limits
 <hr/>

### For `n` species, a network `W`, a set of abiotic variables `E`,
    --

### Community `\(C_k = \{0, 1, 0, 0, ..., 1, 0\}\)`
    --

## `\(P(C_{k,t+1} | C_{l, t}) = f(W, E)\)`

--
 
 ### .right[[Cazelles *et al.*, *Ecography* (2016).](http://doi.org/10.1111/ecog.01714)]

---
 # Conclusion and limits
 <hr/>

### Community-based biogeography: more than occurrence and co-occurrence

--
    ### `\(n\)` species `\(2^n\)` potential communities

--
    ### [What is wrong with co-occurrence data for species embedded in a network?]()

---
 name: style2
 layout: true
 class: center, middle, style2
 ---
 # CO-OCCURRENCE AND ECOLOGICAL NETWORKS
 <hr />
 ### How the properties of ecological networks influence co-occurrence?

---
    name: style1
    layout: true
    class: style1

---
 # Co-occurrence and biotic interactions
 <hr/>

.center[![:scale 100%](./fig/figcooc1.png)]

---
 # Co-occurrence and biotic interactions
 <hr/>

.center[![:scale 100%](./fig/figcooc2.png)]

---
 # Co-occurrence and biotic interactions
 <hr/>

.center[![:scale 100%](./fig/figcooc3.png)]

---
 # Co-occurrence and biotic interactions
 <hr/>

.center[![:scale 100%](./fig/figcooc4.png)]

---
 # Co-occurrence and biotic interactions
 <hr/>

.center[![:scale 100%](./fig/figcooc5.png)]

---
 # Co-occurrence and biotic interactions
 <hr/>

.center[![:scale 100%](./fig/figcooc6.png)]

---
 # Checkerboard distribution (Diamond 1975)
 <hr/>

.center[![:scale 52%](./fig/figCS1.png)]

---
 # Checkerboard distribution (Diamond 1975)
 <hr/>

.center[![:scale 52%](./fig/figCS2.png)]

---
 # Checkerboard distribution (Diamond 1975)
 <hr/>

.center[![:scale 52%](./fig/figCS3.png)]

---
 # What did I do?
 <hr/>

### Checkerboard distrbution in a network context

.column[
 ### Observed `vs` Expected
 
 ### `\(P(X_i,X_j) - P(X_i)P(X_j)\)`
 ]
 .column[

]

---
 # What did I do?
 <hr/>

.column[
 ### Observed `vs` Expected
 
 ### `\(P(X_i,X_j) - P(X_i)P(X_j)\)`
 ]
 .column[
 `\(P(X_i,X_j) - P(X_i)P(X_j)=0\)`

.center[![:scale 32%](./fig/figCS1.png)]
    ]

---
 # What did I do?
 <hr/>

.column[
 ### Observed `vs` Expected
 
 ### `\(P(X_i,X_j) - P(X_i)P(X_j)\)`
 ]
 .column[
 `\(P(X_i,X_j) - P(X_i)P(X_j)<0\)`

.center[![:scale 32%](./fig/figCS2.png)]
    ]

---
 # What did I do?
 <hr/>

.column[
 ### Observed `vs` Expected
 
 ### `\(P(X_i,X_j) - P(X_i)P(X_j)\)`
 ]
 .column[
 `\(P(X_i,X_j) - P(X_i)P(X_j)>0\)`

.center[![:scale 32%](./fig/figCS3.png)]
    ]

---
 # What did I do?
 <hr/>

.column[
 ### Observed `vs` Expected
 
 ### `\(P(X_i,X_j) - P(X_i)P(X_j)\)`

### Simulated co-occurrence (trophic interactions)
 ]
 .column[

.center[![:scale 42%](fig/gtib/figGTIB8.png)]
    ]

---
 # What did I do?
 <hr/>

.column[
 ### Observed `vs` Expected
 
 ### `\(P(X_i,X_j) - P(X_i)P(X_j)\)`

### Simulated co-occurrence (trophic interactions)

### Simulated networks
 ]
 .column[

.center[![:scale 38%](./fig/figweb.png)]
    ]

---
 # Co-occurrence and shortest path
 <hr/>

.center[![:scale 100%](./fig/figsht0.png)]

---
 # Co-occurrence and shortest path
 <hr/>

.center[![:scale 100%](./fig/figsht1.png)]

---
 # Co-occurrence and shortest path
 <hr/>

.center[![:scale 100%](./fig/figsht2.png)]

---
 # Co-occurrence and shortest path
 <hr/>

.center[![:scale 100%](./fig/figsht3.png)]

---
 # Co-occurrence and degree
 <hr/>

.center[![:scale 100%](./fig/figdeg0.png)]

---
 # Co-occurrence and degree
 <hr/>

.center[![:scale 100%](./fig/figdeg1.png)]

---
 # Co-occurrence and degree
 <hr/>

.center[![:scale 100%](./fig/figdeg2.png)]

---
 # Co-occurrence and degree
 <hr/>

.center[![:scale 100%](./fig/figdeg3.png)]

---
 # Conclusion and limits
 <hr/>

## Shortest path    `\( \Longrightarrow \)`    detection

## Degree    `\( \Longrightarrow \)`    detection

--
 
 ### .right[[Cazelles *et al.*, *Theoretical Ecology* (2016).](http://doi.org/10.1007/s12080-015-0281-9)]

---
 # Correlations?

.center[![:scale 100%](./fig/figNew/argument_pred1.png)]

---
 # Correlations?

.center[![:scale 100%](./fig/figNew/argument_pred2.png)]
 ---
 # Correlations?

.center[![:scale 100%](./fig/figNew/argument_pred3.png)]
 ---
 # Correlations?

.center[![:scale 100%](./fig/figNew/argument_pred4.png)]

---
 # Conclusion and limits
 <hr/>

## Shortest path    `\( \Longrightarrow \)`    detection

## Degree    `\( \Longrightarrow \)`    detection

--

## Abiotic variables?

--
    ## Empirical data?

---
 name: style2
 layout: true
 class: center, middle, style2
 ---
 # DEALING WITH REAL DATA SETS
 <hr />
 ### Can we infer interactions from co-occurrence data?

---
    name: style1
    layout: true
    class: style1

---
 # Abiotic variables
 <hr/>

.center[![:scale 90%](./fig/figab1.png)]

---
 # Abiotic variables
 <hr/>

.center[![:scale 90%](./fig/figab2.png)]

---
 # Integrating abiotic variables (E)
 <hr/>

# `\(P(X_{i,E},X_{j,E})\)` `vs`  `\(P(X_{i,E})P(X_{j,E})\)`

---
 # Integrating abiotic variables (E)
 <hr/>

.center[![:scale 100%](./fig/figshtenv1.png)]

---
 # Integrating abiotic variables (E)
 <hr/>

.center[![:scale 100%](./fig/figshtenv2.png)]

---
 # What did I do?
 <hr/>

### 4 data sets, **know interactions**
    --

### `\(P(X_{i,E},X_{j,E})\)` `vs`  `\(P(X_{i,E})P(X_{j,E})\)`
    --

### SDMs to assign presence probabilities given abiotic context
    --

### Detection once abiotic variables integrated

---
 # Kopelke *et al.*, *Ecology* (2017)
 <hr/>

.column[
      .center[![:scale 87%](./fig/mapsal1.png)]
    ]

.column[
      .center[![:scale 25%](./img/cool.png)]
    ]

???
    mettre les noms

---
 # Kopelke *et al.*, *Ecology* (2017)
 <hr/>

.column[
      .center[![:scale 87%](./fig/mapsal2.png)]
    ]

.column[
      .center[![:scale 25%](./img/cool.png)]
    ]

???
    symphyte / ovipositeur

---
 # Three trophic levels network
 <hr/>

# Willow Galler Parasitoids
 # 52         96         126     species

--
 
 # Willow Galler

--
 # Galler Parasitoids

---
 # Willow Galler - Shortest path
 <hr/>

![:scale 100%](./fig/figSG0.png)

---
 # Willow Galler - Shortest path
 <hr/>

![:scale 100%](./fig/figSG1.png)

---
 # Willow Galler - Shortest path
 <hr/>

![:scale 100%](./fig/figSG2.png)

---
 # Willow Galler - Shortest path
 <hr/>

![:scale 100%](./fig/figSG3.png)

---
 # Willow Galler - Shortest path
 <hr/>

![:scale 100%](./fig/figSG4.png)

---
 # Willow Galler - `\(P(X_{i,E},X_{j,E})\)`
 <hr/>

![:scale 100%](./fig/figSG5.png)

---
 # Galler Parasitoids - `\(P(X_{i,E},X_{j,E})\)`
 <hr/> 
 ![:scale 100%](./fig/figGP5.png)

---
 # Co-occurrence and shortest path
 <hr/>

.center[![:scale 100%](./fig/figsht3.png)]

---
 # Willow Galler `vs` Galler Parasitoids
 <hr/>

.center[![:scale 53%](./fig/figSG5.png)]
    .center[![:scale 53%](./fig/figGP5.png)]

---
 # Willow Galler `vs` Galler Parasitoids
 <hr/>

.center[![:scale 53%](./fig/figSG6.png)]
    .center[![:scale 53%](./fig/figGP6.png)]

---
 # Willow Galler `vs` Galler Parasitoids
 <hr/>

.center[![:scale 53%](./fig/figSG7.png)]
    .center[![:scale 53%](./fig/figGP7.png)]

---
 # Co-occurrence and degree
 <hr/>

.center[![:scale 100%](./fig/figdeg3.png)]

---
 # Conclusion
 <hr/>

## Shortest path    `\( \Longrightarrow \)`    detection

--
 ## Degree    `\( \Longrightarrow \)`    detection

--
    ## Using empirical data sets.

--
 
 ### .right[[Cazelles *et al.*, *in prep* (2018).]()]

---
 name: style2
 layout: true
 class: center, middle, style2
 ---
 # Concluding remarks and perspectives
 <hr/>
 ### Towards better predictions?

---
    name: style1
    layout: true
    class: style1
    ---

# Answers
 <hr/>

## 1. How to integrate biotic interactions into distribution models?

--
    ### `\( \Longrightarrow \)` think at the community-scale

--
    ### `\( \Longrightarrow \)` build Network Distribution Models (NDMs)

---
 # Answers
 <hr/>

## 2. How the properties of ecological networks influence co-occurrence?
    --

### `\( \Longrightarrow \)` shortest path    `\( \Longrightarrow \)`    detection 
 ### `\( \Longrightarrow \)` degree    `\( \Longrightarrow \)`    detection

### We may not need to consider the whole network

---
 # Answers
 <hr/>

## 3. Can we infer interactions from co-occurrence data?
    --

### `\( \Longrightarrow \)` depends on the network properties

### `\( \Longrightarrow \)` depends on the spatial scale (literature)

???
    variations des interactions

---
 # Is it correct?
 <hr/>

.center[![:scale 90%](./fig/figshtenv2.png)]

## .center[INTERDEPENDENCY]

---
 # Towards mechanistic NDMs?
 <hr/>

## Better than JSDMs? (more information)

## Network Distribution Models (NDMs)

## Network `\(W\)`, abiotic variables `\(E\)`

## How to predict interactions?

---
 # Towards mechanistic NDMs?
 <hr/>

.center[![:scale 100%](./fig/figLast1.png)]

---
 # Towards mechanistic NDMs?
 <hr/>

.center[![:scale 100%](./fig/figLast2.png)]

---
 # Towards mechanistic NDMs?
 <hr/>

.center[![:scale 100%](./fig/figLast3.png)]

---
 # Towards mechanistic NDMs?
 <hr/>

.center[![:scale 100%](./fig/figLast4.png)]

---
 # Towards mechanistic NDMs?
 <hr/>

.center[![:scale 100%](./fig/figLast5.png)]

---
 # Towards mechanistic NDMs?
 <hr/>

### 1. **Building a biogeography of ecological networks**

![:fontawe](file-text) [Galiana, N., *et al.* (2018). The spatial scaling of species interaction networks. *Nature Ecology & Evolution*.](https://www.nature.com/articles/s41559-018-0517-3)

### 2. **[Building an energetic framework](https://kevcaz.github.io/talkBES2017/#1)**

---
 name: style2
 layout: true
 class: center, middle, style2
 ---
 # THANK YOU
 <hr />
 ### Merci

---
    name: style2
    layout: true
    class: class: center, middle, style2
    ---

## What am I doing these days?
 <hr />
 ### Science, what else?

---
    name: style1
    layout: true
    ---

# Homogenization
 <hr>

![](img/homogen.png)

---

# Rewiring + seasonality in food webs
 <hr>

### [Nature rewires in a changing world](https://peerj.com/preprints/27187/) *Nature Eco Evo*

---
 # Iscoscape: the push for provenance
 <hr>

### [Fighting noise with dimensionality](https://kevcaz.github.io/fightingNoise/#1)

---
 name: style2
 layout: true
 class: center, middle, style2
 ---
 # THANK YOU
 <hr />
 ### Merci

---
    name: style1
    layout: true
    class: center, style1

---

---
 # Urgently needed
 <hr/>

## Interactions ++ predictions easier

### Berlow *et al*, 2009, *PNAS*
    --

## WWF, Living Planet Report 2016 [](http://awsassets.wwf.ca/downloads/wwf_living_planet_report_2016___risk_and_resilience_in_a_new_era.pdf)

>  If current trends continue to 2020
    > vertebrate populations may decline by an
    > average of 67 per cent compared to 1970.