Use this URL to cite or link to this record in EThOS:
Title: Stochastic models of animal movement and habitat selection
Author: Michelot, Théo
ISNI:       0000 0004 7960 376X
Awarding Body: University of Sheffield
Current Institution: University of Sheffield
Date of Award: 2019
Availability of Full Text:
Access from EThOS:
Access from Institution:
The analysis of animal movement reveals important features of habitat preferences and behaviours, and informs environmental conservation decisions. In this thesis, we present new statistical methods, to tackle the problem of scale dependence in models of animal movement. The inferences obtained from most existing approaches are tied to a particular spatio-temporal scale, which makes the interpretation and comparison of results difficult. We first focus on models of habitat selection, which combine tracking data and environmental data, to understand the drivers of animal movement. The two most popular approaches describe habitat selection on two different scales, and their parameters have different interpretations. We propose a time series approach to integrate local and global habitat selection. We explain how stochastic processes with known stationary distributions can be used, to describe both the short-term transition density and the long-term equilibrium distribution of the movement. The proposed approach captures both the short-term and long-term habitat selection. We suggest using Markov chain Monte Carlo (MCMC) algorithms to model animal movement. A MCMC algorithm describes transition rules, which lead to a limiting distribution: its target distribution. We also suggest the Langevin diffusion process as a continuous-time model of movement with known stationary distribution. We describe methods of estimation, to obtain habitat selection and movement parameters from tracking data. We then turn to the problem of the time formulation in models of animal movement and behaviour. Most widely-used models describe movement in discrete time, and their results are tied to the time scale of the observed data. We extend a popular continuous-time model of movement, to include behavioural heterogeneity. The approach can be used to identify behavioural phases from movement data collected at irregular intervals, and with measurement error. We describe a framework of Bayesian inference, to estimate movement parameters and behavioural phases from tracking data.
Supervisor: Blackwell, Paul G. Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available