#PhDPosition in Ecological #DataScience in our group. For details, see www.uni-regensburg.de/universitaet... #MachineLearning #AI #DeepLearning #Statistics #Ecology #AcademicJobs
Posts by Martin Philippe-Lesaffre
🏝️ Saviez-vous que 3/4 des extinctions récentes ont eu lieu sur des îles ?
#FRB-Cesab #Rivage propose un cadre inédit pour évaluer cette vulnérabilité. Leur objectif : replacer les îles au cœur des priorités de conservation.
📌 bit.ly/4mrUBil
@celinebellard.bsky.social @claramarino.bsky.social
📣 We see this mapping as a step toward a more comprehensive threat landscape.
Thanks to all co-authors : @anabenlop.bsky.social @iagoferreiro.bsky.social @brodieecology.bsky.social Laura Maeso-Pueyo and Dominik Schüßler.
and to our academic support : @mncn-csic.bsky.social @mncn-bgcg.bsky.social
🌎🌍🌏Hotspots and refuges:
🔥 Southeast Asia, West Africa, Atlantic Forest
🛡️ Remote areas of Borneo, Central Africa, Western Amazon
📊 Findings:
🔻 Only ~11% of tropical forest shows low pressure (HP < 0.5)
🔻 ~9% qualifies as high-pressure hotspots (HP > 0.9)
🔻 Net global increase from 2000–2015, driven by increased access (notably Amazon, Southeast Asia)
📈 Model performance:
High predictive performance + meaningful relationships between hunting probability and key drivers of hunting pressure (HP):
🔻Distance to the first human settlement
🔻Forest patch size
🔻Socioeconomic context
🔻Protected area
👨💻👩💻 Our approach:
We built a species-agnostic machine learning approach using ecological and socio-economic :
🔻 2,463 geo-referenced sites (hunted & non-hunted)
🔻 Random forest classification to predict hunting prob.
🔻 Predictors: accessibility, habitat quality, human context
🔻 1 km² resolution
💡Why a global map of hunting pressure?
Hunting is a leading cause of biodiversity loss, especially for tropical mammals and birds.
Yet, until now, we lacked spatially explicit, standardized metric across the tropics.
Previous efforts were species-specific, regional or relied on IUCN-based proxies.
🖨️ New preprint out !
We present the first global, high-resolution, spatiotemporal maps of hunting pressure across tropical forests 🦜🐒🌴.
This long-standing blind spot in biodiversity threat mapping is now filled 😎:
ecoevorxiv.org/repository/v...
New paper out! We present a framework for quantifying the vulnerability of terrestrial insular biota to multiple threats. This is the first output of the #RIVAGE project, funded by @frbiodiv.bsky.social & brilliantly led by @celinebellard.bsky.social.
peercommunityjournal.org/articles/10....
⚠️Some birds rarely recorded as cat prey might still be highly vulnerable based on their traits alone.
Thanks to my co-author Elsa Bonnaud and to @univparissaclay.bsky.social for supporting this research.
🌍 When assemblages of species are similar, domestic cats consistently target birds with similar traits —size, behavior, and range matter everywhere.
🪶The trait-based approach offers the possibility of predicting the vulnerability of birds when data is insufficient.
Our paper on 🐈 vs. 🐦 , an overview 🔐:
Ever wonder how domestic cats choose their prey? We combined citizen science and machine learning to find out what makes birds attractive to cats. Small size, generalist habits, and large range put birds at higher risk!
www.sciencedirect.com/science/arti...
I'm really happy to see that more and more articles on ecology are highlighting the fact that we really need to adopt more causal approaches or at least try !
🪱 Putting Earthworm conservation on the map!
Check out the last publication of our #CESAB LANDWORM and IMPACT groups by @sylvaingrd.bsky.social et al.
👉 bit.ly/4ij7sTg
🔑 They are soil keystones & face global threats, demanding broad-scale, multifaceted diversity indicators for conservation.
🧪🌐🌍🦤
Many thanks to all the co-authors, (@conservbytes.bsky.social and @franckcourchamp.bsky.social already on Bluesky) and to the Université Paris-Saclay for making this study possible.
It also improves knowledge of the species most likely to be depredated by cats on mainlands, something that has been little studied outside Australia. We hope these results will help to better understand the potential impacts of free-ranging cats on mainlands.
The predation pattern showed a shift in favor of species with higher body mass, compared to Europe and North America. Our study showed that prey-predator relationships can change even on a macro-ecological scale, but the co-evolutionary context seems to strongly mediate this.
We observed that the likelihood of being depredated by cats depended on traits and phylogenies, with some species being more likely to be depredated than others, and this dependence is similar between continents, with the exception of Australian mammals.
Combining observed and predicted cat diet data from three continents, Australia, Europe and North America, we compared prey and non-prey species groups within and between continents. These comparisons were made on the basis of species characteristics and phylogenies.