Aurélien Boyé
@aurelien-boye.bsky.social
Benthic ecologist at @ifremer.bsky.social🐬
• Using numerical ecology to understand and predict marine coastal biodiversity, focusing on the role of foundation species like seagrass 🌱 and maerl 🪸
• Trait-based ecology, community trajectory, ML, Rstats, ☕
• Using numerical ecology to understand and predict marine coastal biodiversity, focusing on the role of foundation species like seagrass 🌱 and maerl 🪸
• Trait-based ecology, community trajectory, ML, Rstats, ☕
Nice ! Can I be added too ? Thanks !
February 19, 2025 at 1:27 PM
Nice ! Can I be added too ? Thanks !
Reposted by Aurélien Boyé
🌍 Understanding habitat states & transitions is key for conservation!
🏝️ Helps track ecosystem shifts from climate change
🛑 Detects early warning signs of degradation (e.g., kelp loss → urchin barrens).
🏝️ Helps track ecosystem shifts from climate change
🛑 Detects early warning signs of degradation (e.g., kelp loss → urchin barrens).
February 13, 2025 at 10:00 AM
🌍 Understanding habitat states & transitions is key for conservation!
🏝️ Helps track ecosystem shifts from climate change
🛑 Detects early warning signs of degradation (e.g., kelp loss → urchin barrens).
🏝️ Helps track ecosystem shifts from climate change
🛑 Detects early warning signs of degradation (e.g., kelp loss → urchin barrens).
Reposted by Aurélien Boyé
🌍 With our data-driven standardized way to classify & monitor marine habitats, what did we find?
🏝️ 17 distinct habitat states, including kelp forests, coral reefs & transitional zones
📉 Our pipeline captures fine-scale ecological changes over time
🏝️ 17 distinct habitat states, including kelp forests, coral reefs & transitional zones
📉 Our pipeline captures fine-scale ecological changes over time
February 13, 2025 at 10:00 AM
🌍 With our data-driven standardized way to classify & monitor marine habitats, what did we find?
🏝️ 17 distinct habitat states, including kelp forests, coral reefs & transitional zones
📉 Our pipeline captures fine-scale ecological changes over time
🏝️ 17 distinct habitat states, including kelp forests, coral reefs & transitional zones
📉 Our pipeline captures fine-scale ecological changes over time
Reposted by Aurélien Boyé
This study wouldn’t be possible without citizen scientists! 🌍🌊
👩🔬 Reef Life Survey divers collected 6554 transects worldwide
📸 Standardized photoquadrat sampling ensured consistency
📊 It enabled us to track habitat transitions over time
👩🔬 Reef Life Survey divers collected 6554 transects worldwide
📸 Standardized photoquadrat sampling ensured consistency
📊 It enabled us to track habitat transitions over time
February 13, 2025 at 10:00 AM
This study wouldn’t be possible without citizen scientists! 🌍🌊
👩🔬 Reef Life Survey divers collected 6554 transects worldwide
📸 Standardized photoquadrat sampling ensured consistency
📊 It enabled us to track habitat transitions over time
👩🔬 Reef Life Survey divers collected 6554 transects worldwide
📸 Standardized photoquadrat sampling ensured consistency
📊 It enabled us to track habitat transitions over time
Reposted by Aurélien Boyé
Machine learning models like UMAP & HDBSCAN are powerful, but they’re often hard to interpret.
So, we used SHAP to:
📌 Identify which features drive cluster formation
📌 Uncover non-linear interactions in benthic ecosystems
📌 Improve ecological interpretability of results
So, we used SHAP to:
📌 Identify which features drive cluster formation
📌 Uncover non-linear interactions in benthic ecosystems
📌 Improve ecological interpretability of results
February 13, 2025 at 10:00 AM
Machine learning models like UMAP & HDBSCAN are powerful, but they’re often hard to interpret.
So, we used SHAP to:
📌 Identify which features drive cluster formation
📌 Uncover non-linear interactions in benthic ecosystems
📌 Improve ecological interpretability of results
So, we used SHAP to:
📌 Identify which features drive cluster formation
📌 Uncover non-linear interactions in benthic ecosystems
📌 Improve ecological interpretability of results
Reposted by Aurélien Boyé
We used better methods: UMAP (for dimension reduction) & HDBSCAN (for clustering).
✅ Captures complex ecological structures
✅ Finds clusters of varying shapes & sizes
✅ Filters noise instead of forcing data into clusters
✅ Captures complex ecological structures
✅ Finds clusters of varying shapes & sizes
✅ Filters noise instead of forcing data into clusters
February 13, 2025 at 10:00 AM
We used better methods: UMAP (for dimension reduction) & HDBSCAN (for clustering).
✅ Captures complex ecological structures
✅ Finds clusters of varying shapes & sizes
✅ Filters noise instead of forcing data into clusters
✅ Captures complex ecological structures
✅ Finds clusters of varying shapes & sizes
✅ Filters noise instead of forcing data into clusters
Reposted by Aurélien Boyé
Clustering is widely used in ecology to define community types, habitat states & biodiversity patterns. But common methods used in ecology have major drawbacks. (See this excellent video of @JohnDataHealy
youtu.be/dGsxd67IFiU?...)
youtu.be/dGsxd67IFiU?...)
HDBSCAN, Fast Density Based Clustering, the How and the Why - John Healy
YouTube video by PyData
youtu.be
February 13, 2025 at 10:00 AM
Clustering is widely used in ecology to define community types, habitat states & biodiversity patterns. But common methods used in ecology have major drawbacks. (See this excellent video of @JohnDataHealy
youtu.be/dGsxd67IFiU?...)
youtu.be/dGsxd67IFiU?...)
Could you also add me please ? Thanks for creating this !
November 24, 2024 at 11:45 AM
Could you also add me please ? Thanks for creating this !
I would love to be added ! Thanks for this list !
November 24, 2024 at 10:19 AM
I would love to be added ! Thanks for this list !
