| Summary: | In my thesis, I develop mechanistic models of space use based on animal movement, to understand and to predict population distribution in heterogeneous and dynamic landscapes. Used and developed methodologies couple mathematical modelling of the spatio-temporal dynamics of animal movement together with statistical analysis of simulated and empirical movement datasets. In my first chapter, I proceed in a series of simulations to clarify how many clusters are needed when using generalized estimating equations to correctly account for the correlation in movement data and to obtain robust inference on habitat selection. My simulations reveal that 30 independent individuals, each assigned to a cluster, are sufficient to avoid biased evaluation of the uncertainty on habitat selection along movement in heterogeneous environments. When less than 30 individuals are available, destructive sampling can be used but solely when temporal correlation is present and inter-individual heterogeneity is low in the data. In my second chapter, I develop a statistical movement model that allows to identify successive behavioral phases (e.g., foraging phase, inter-patch movement) together with behavior-specific habitat selection parameters, over the whole population and using temporally irregular data. Analysis of simulated and empirical movement data from three large herbivores including plains bison (Bison bison bison), mule deer (Odocoileus hemionus) and plains zebra (Equus quagga) show the robustness and the high predictive capacity of the model. This statistical tool is also flexible since I assess multiple ecological processes from those datasets such as foraging behavior, migratory behavior or prey-predator interactions. In addition, I show how accounting for behavioral phases in habitat selection analysis is crucial to correctly characterize habitat selection along animal movement. In my third chapter, I develop a mathematical framework to couple movement of individuals among a network of resource patches with residency time in ...
|
|---|