Integrating heterogeneous biodiversity data to understand species spatial dynamics

Title: Integrating heterogeneous biodiversity data to understand species spatial dynamics
 

Abstract: Large citizen science platforms (e.g. Pl@ntNet, iNaturalist, eBird) have recently brought a massive and continuous stream of accessible biodiversity observations with wide coverage, and have opened up new perspectives in biogeography and for our understanding of recent ecosystem changes. However, the analysis of this presence-only data is often difficult due to their heterogeneity and the many observation biases that notably distort our vision of species distribution in space and time. Statistical modeling of observation biases and the integration of presence only data with more standardized observations can nevertheless correct this vision. In my research, I use statistical modeling and machine learning to model and integrate heterogeneous data. Thereby, I try to estimate and anticipate the effects of human pressure on ecosystems, such as land use changes or biological invasions, and in gaining insights for conservation planning. I'll present an overview of my main research themes over the past years, including (i) methods to correct sampling biases when estimating species distribution in space and time from opportunistic presence-only data, (ii) using such data to reconstruct local tetrapod food webs and analyse their structural variations related to land use intensification in Europe, and (iii) estimating mechanistic spatial dynamic models from heterogeneous observations to reconstruct and understand biological invasion trajectories.