Statistical methods and software tools to analyze and infer ecological networks and process multi-species data
Published:
My thesis aims at providing tools to improve our understanding of ecological communities. To this end, I tackled diverse research questions and methodologies.
In a first axis, I use multivariate methods to analyze bird-plants frugivory networks. These methods provide insights into trait matching and niche-based processes. Case studies suggest that there is a signal of trait matching, but that it is weak in frugivory networks.
In a second axis, I use point processes of the multivariate Hawkes process family to infer interspecific interactions from camera trap data. I test these methods using simulations, and a case study on African mammals. Simulations suggest that under certain conditions, multivariate Hawkes processes can be used to analyze camera trap data. The case study highlights expected patterns of attraction-avoidance between lion, impala, kudu, wildebeest and zebra.
In a third axis, I present two software tools designed during my thesis to analyze camera trap data ({camtrapviz} and {standardizeSnapshot}).
In the discussion of the manuscript, I also share some thoughts on open science and reproducibility, the epistemology of ecology and my experience of research during my PhD.
Recommended citation: Nicvert, L. (2024). Statistical methods and software tools to analyze and infer ecological networks and process multi-species data [PhD thesis, Université Claude Bernard - Lyon I]. https://theses.hal.science/tel-04751639
Download Paper | Download Slides | Download Bibtex