EcoGraph

Species distribution models lie at the heart of many ecological applications, such as conservation, climate change research or invasive species monitoring. Knowledge on where species occur or can occur in space is often crucial, but modelling species distributions is complex due to the intricate nature of the task. Despite strong developments, including the integration of deep learning methods, most of the current models still operate on gridded spatial data, and often keep struggling with species interactions, imbalanced, biased data and spatial complexity.

In the EcoGraph project, we will address some of these aspects by using networks to represent and process spatial structures and species communities. Starting off from simulations, we will gradually shift to existing multispecies datasets to apply our new approaches and compare them with existing benchmarks.

Researchers: Reinout Vandenabeele

Funding: FWO PhD grant