From macroecology to biosecurity through species distribution modeling

class: hide-count, full, hide_logo, center, top, inverse
exclude: false
.titrejp[From species distribution modeling to biosecurity] 
.bigjp[Jean-Pierre Rossi] 
.center[<img src="im1/fig3d2.png" title="" alt="" width="400" />]
.small60[2023, September 19th]

---
class: hide_logo, inverse, left, middle
exclude: false
# 1) Introducing Species Distribution Models (SDMs)

---
exclude: false
# SDM in a nutshell 
.center[<img src="im1/sdm/sdm1.png" title="" alt="" width="800"/>]

---
exclude: false
# SDM in a nutshell 
.center[<img src="im1/sdm/sdm2.png" title="" alt="" width="800"/>]

---
exclude: false
# SDM in a nutshell 
.center[<img src="im1/sdm/sdm3.png" title="" alt="" width="800"/>]

---
exclude: false
# SDM in a nutshell 
.center[<img src="im1/sdm/sdm4.png" title="" alt="" width="800"/>]

---
exclude: false
# SDM in a nutshell 
.center[<img src="im1/sdm/sdm5.png" title="" alt="" width="800"/>]

---
exclude: false
# SDM in a nutshell 
.center[<img src="im1/sdm/sdm6.png" title="" alt="" width="800"/>]

---
class: hide_logo, inverse, left, middle
exclude: false
## Examples

---
exclude: false
<img class="logo" src="im1/Hv-nom.png" alt="logo" width="180">
# SDM outputs: environemental suitability
Index of suitability comprised between 0 and 1
.center[<img src="im1/Hv1.png" title="" alt="" width="900"/>]
.conteneur[Rossi & Rasplus, 2023]

---
exclude: false
<img class="logo" src="im1/Hv-nom.png" alt="logo" width="180">
# Reclassed environemental suitability
Qualitative classes for better readability
.center[<img src="im1/Hv2.png" title="" alt="" width="900"/>]
.conteneur[Rossi & Rasplus, 2023]

---
exclude: false
<img class="logo" src="im1/Hv-nom.png" alt="logo" width="180">
# Projecting outside species native range
Europe, current climate
.center[<img src="im1/Hv3.png" title="" alt="" width="450"/>]
.conteneur[Rossi & Rasplus, 2023]

---
exclude: false
<img class="logo" src="im1/Hv-nom.png" alt="logo" width="180">
# Projecting according to future climate conditions
Europe, future conditions (2081–2100), ssp245
.center[<img src="im1/Hv4.png" title="" alt="" width="450"/>]
.conteneur[Rossi & Rasplus, 2023]

---
exclude: false
<img class="logo" src="im1/Ld-nom.png" alt="logo" width="180">
# Response curves: understanding species-environment relationships
The optimal range of the temperature (colder month) is between ca. -12 and -2 °C
.center[<img src="im1/Ld1.png" title="" alt="" width="500"/>]
.conteneur[Chartois et al in review]

---
exclude: false
<img class="logo" src="im1/Ld-nom.png" alt="logo" width="180">
# Partial dependences to map limiting factors
The species response to the variables can be mapped in the geographical space 
Areas of high partial dependence `\(\rightarrow\)` areas where the variable is favourable
.center[<img src="im1/Ld2.png" title="" alt="" width="900"/>]
.conteneur[Chartois et al in review]

---
exclude: false
# Data and data problems &#128556;

- <ins>Presence data</ins>: the species is present in the form of viable populations

- <ins>Absence data</ins>: the species is absent because it cannot maintain in the long term (true absences)

- <ins>Pseudo-absences</ins>: absence points generated in areas where the species is considered to be absent. **Assumption** : if the species is absent, it is due to environmental (climate) unsuitability not to dispersal contraints (beware of barriers)

---
exclude: false
<img class="logo" src="im1/Ld-nom.png" alt="logo" width="180">
# Data and data problems: pseudo-absences
- We cannot generate pseudo-absences in the US because invasion is currently underway
- We can can generate pseudo-absences at north and south of the native range which shape is well documented
.center[<img src="im1/Ld3.png" title="" alt="" width="900"/>]
.conteneur[Chartois et al in review]

---
exclude: false
<img class="logo" src="im1/Ld-nom.png" alt="logo" width="180">
# Data and data problems: sampling bias
Sampling bias occurs when data availability changes systematically in the geographical space 
Native range often under-sampled while newly invaded areas often over-sampled 
Solution: data thinning
.center[<img src="im1/Ld4.png" title="" alt="" width="600"/>]
.conteneur[Rossi unpublished]

---
exclude: false

# The SDM ecosystem: so diverse !

.pull-left[
**Data type and volume determine the method**
 .conteneur[Guisan et al 2017]
- Presence/absence data: GLM, various regression tree methods,…
- Presence but no absence:  profil methods
- Presence, no absence + background points: Maxent
 ]

.pull-right[
 .center[<img src="im1/Guisan.png" title="" alt="" width="300"/>]
]

---
class: hide_logo, inverse, left, middle
exclude: false
# 2) Better understand invasions and improve biosecurity

---
exclude: false
<img class="logo" src="im1/Tp-nom.png" alt="logo" width="180">
# Climate-based constraints and biotic interactions
 
.center[<img src="im1/Tp1.png" title="" alt="" width="800"/>]
.conteneur[Rossi et al in prep]

---
exclude: false
<img class="logo" src="im1/Tp-nom.png" alt="logo" width="180">
# Climate-based constraints and biotic interactions
 
.center[<img src="im1/Tp2.png" title="" alt="" width="800"/>]
.conteneur[Rossi et al in prep]

---
exclude: false
<img class="logo" src="im1/Tp-nom.png" alt="logo" width="180">
# Climate-based constraints and biotic interactions
 
.center[<img src="im1/Tp4.png" title="" alt="" width="800"/>]
.conteneur[Rossi et al in prep]

---
exclude: false
<img class="logo" src="im1/Tp-nom.png" alt="logo" width="180">
# Climate-based constraints and biotic interactions
Biotic interactions should not be overlooked!
.center[<img src="im1/Tp5.png" title="" alt="" width="800"/>]
.conteneur[Rossi et al in prep]

---
exclude: false
<img class="logo" src="im1/Xf-nom.png" alt="logo" width="180">
# Potential range can help design pre-emptive studies
Knowing the geography of risk help being prepared in case of incursion
.center[<img src="im1/Xf1.png" title="" alt="" width="700"/>]
.conteneur[Godefroid et al 2019]

---
exclude: false
<img class="logo" src="im1/Wa-nom.png" alt="logo" width="180">
# Population level matters
- Origin: tropical south america
- Invasion of Israel: cold and dry conditions
.center[<img src="im1/Wa1.png" title="" alt="" width="400"/>]
.conteneur[Rey et al 2012]

---
exclude: false
<img class="logo" src="im1/Wa-nom.png" alt="logo" width="180">
# Population level matters
- Modeling the niche of the invasive population and projection onto the native range
- Points toward the origin of the invaders (confirmed by genetic studies)
.center[<img src="im1/Wa2.png" title="" alt="" width="400"/>]
.conteneur[Rey et al 2012]

---
exclude: false
<img class="logo" src="im1/Dp-nom.png" alt="logo" width="180">
# Infra-specific level matters too
- Lineages have different potential ranges
- The global model hides this level of detail potentially important for biosecurity
.center[<img src="im1/Dp1.png" title="" alt="" width="900"/>]
.conteneur[Godefroid et al 2016]

---
exclude: false
<img class="logo" src="im1/Xc-nom.png" alt="logo" width="180">
# Infra-specific level matters too
The "**crassiusculus affair**"

Urvois et al 2021: SDM at species level fails .titrejp[&#128560;]

Urvois et al 2023: genetic characterization of 2 genetic lineages .titrejp[&#128512;]

Urvois et al in review: SDM at lineage level works! .titrejp[&#128526;]

.center[.content-box-green[Species level is not always the level at which we should work!]]

---
exclude: false
<img class="logo" src="im1/Xc-nom.png" alt="logo" width="180">
# Taxonomic level matters too
Any divergence in climatic niches? yes

Hellingers’ *I* = 0.310 (p < 0.01, 1000 randomizations) ; niche comparison based on Warren et al. (2008)

.center[<img src="im1/Xc2.png" title="" alt="" width="900"/>]

.conteneur[Urvois et al in review]

---
exclude: false
<img class="logo" src="im1/Xc-nom.png" alt="logo" width="180">
# Taxonomic level matters too
Potential distribution of lineage #1 (absent from Europe)
.center[<img src="im1/Xc3.png" title="" alt="" width="900"/>]
.conteneur[Urvois et al in review]

---
exclude: false
<img class="logo" src="im1/Xc-nom.png" alt="logo" width="180">
# Taxonomic level matters too
Potential distribution of lineage #2 (invader in Europe)
.center[<img src="im1/Xc4.png" title="" alt="" width="900"/>]
.conteneur[Urvois et al in review]

---
exclude: true
<img class="logo" src="im1/Xc-nom.png" alt="logo" width="180">
# Taxonomic level matters too
.small80[Europe is climatically suitable for both lineages but only lineage #2 is present 
Hypothesis: climate change has improved suitability for lineage #2 but not #1 ? 
- Projection of the model using past climate data
- Climate have been favourable for both lineages since early 1900
- Lineage 1 is absent from Europe by chance or for a question of propagule pressure rather than because of climate]
.center[<img src="im1/Xc5.png" title="" alt="" width="650"/>]
.conteneur[Urvois et al in review]

---
exclude: false
<img class="logo" src="im1/Xc-nom.png" alt="logo" width="180">
# Predicting biorisks in the future (lineage #2)
Current potential distribution
.center[<img src="im1/Xc6.png" title="" alt="" width="700"/>]
.conteneur[Urvois et al in review]

---
exclude: false
<img class="logo" src="im1/Xc-nom.png" alt="logo" width="180">
 
# Predicting biorisks in the future (2041-2070)
Range shrinking in the south, expansion in the north
.center[<img src="im1/Xc7.png" title="" alt="" width="700"/>]
.conteneur[Urvois et al in review]

---
exclude: false
<img class="logo" src="im1/Xc-nom.png" alt="logo" width="180">
# Predicting biorisks in the future (2041-2070)
Important commercial seaports will become climatically suitable
.center[<img src="im1/Xc8.png" title="" alt="" width="700"/>]
.conteneur[Urvois et al in review]

---
class: hide_logo, inverse, left, middle
exclude: false
# Conclusions

---
class: hide_logo, inverse, left, middle
exclude: false

Species distribution models are useful to
  * Identify and map environmental drivers of species distribution
  * Assess potential species range 
  * Evaluate the effects of climate change (in absence of significant adaptation)
  * Identify areas at risk and help managers to set up accurate surveillance plans

.black[&#9888;&#65039; But species distribution models are not silver bullets
  * They lack a proper account of species adaptation
  * They are only as good as the data used to calibrate them (often poor)
  * Huge and confusing literature with many papers of poor statistical quality]

---
class: hide_logo, inverse, left, middle
exclude: false

&#9888;&#65039; But species distribution models are not silver bullets
  * They lack a proper account of species adaptation
  * They are only as good as the data used to calibrate them (sometimes poor)
  * Huge literature with many papers of poor statistical quality

---
class: hide_logo, inverse, left, middle
exclude: false

What do we need to use SDM?
- As many occurrence points as possible
- Absence data when possible
- Information about infra-specific genetic structures if available

Which target species?
- Species representing a danger to forest health
- Select a pool of american, asian and european bark beetles

---
class: hide_logo, left, middle
exclude: false

## References

.small50[
- Chartois, M., Fried, G. & Rossi J-P (in review). Climatic suitability and host plant availability are favorable to the establishment of *Lycorma delicatula* in Europe. Agriocultural and Forest Entomology.

- Guisan, A., Thuiller, W., Zimmermann, N.E. 2017. Habitat suitability and distribution models with applications in R, Ecology, biodiversity and conservation. Cambridge University Press.

- Godefroid, M., Rasplus, J.-Y., Rossi, J.-P. 2016. Is phylogeography helpful for invasive species risk assessment? The case study of the bark beetle genus Dendroctonus. Ecography 39, 1197–1209. doi.org/10.1111/ecog.01474

- Godefroid, M., Cruaud, A., Streito, J.-C., Rasplus, J.-Y., Rossi, J.-P. 2019. *Xylella fastidiosa*: climate suitability of European continent. Sci Rep 9, 8844. doi.org/10.1038/s41598-019-45365-y

- Rey, O., Estoup, A., Vonshak, M., Loiseau, A., Blanchet, S., Calcaterra, L., Chifflet, L., Rossi, J.-P., Kergoat, G.J., Foucaud, J., Orivel, J., Leponce, M., Schultz, T., Facon, B. 2012. Where do adaptive shifts occur during invasion? A multidisciplinary approach to unravelling cold adaptation in a tropical ant species invading the Mediterranean area. Ecology Letters 15, 1266–1275. doi.org/10.1111/j.1461-0248.2012.01849.x

- Rossi, J.-P. & Rasplus, J.-Y., 2023. Climate change and the potential distribution of the glassy-winged sharpshooter (*Homalodisca vitripennis*), an insect vector of *Xylella fastidiosa*. Science of The Total Environment 860, 160375. doi.org/10.1016/j.scitotenv.2022.160375

- Urvois, T., Auger-Rozenberg, M.A., Roques, A., Rossi, J.-P., Kerdelhue, C. 2021. Climate change impact on the potential geographical distribution of two invading *Xylosandrus* ambrosia beetles. Sci Rep 11, 1339. https://doi.org/10.1038/s41598-020-80157-9

- Urvois T., Perrier C., Roques A., Sauné L., Courtin C., Kajimura H., Hulcr J., Cognato A., Auger-Rozenberg M.-A. & Kerdelhué C. 2023. The worldwide invasion history of a pest ambrosia beetle inferred using population genomics. Molecular Ecology. dx.doi.org/10.1111/mec.16993

- Urvois, T., Auger‑Rozenberg, M.A., Roques, A., Kerdelhué, C. & Rossi J.-P. (in review). Intraspecific niche models for the invasive ambrosia beetle Xylosandrus crassiusculus suggest contrasted responses to climate change. Oecologia.

- Warren, D.L., Glor, R.E. & Turelli, M. 2008. Environmental niche equivalency versus conservatism: quantitative approaches to niche evolution. Evolution 62, 2868–2883. doi.org/10.1111/j.1558-5646.2008.00482.x

]