Analysing animal movement in the environment

Abstract

Movement is a crucial part in the life of animals. It determines their locations and their potential interactions. The interest in studying movement has increased in recent years. This increase is associated technological and methodological developments.In this thesis, methods to study movement are investigated and used to predict the location and evolution of migratory routes. Many other methods to study movement data rely on temporal independence of the observations. The temporal correlation however contains useful information and therefore should not be ignored or removed. In this thesis, I incorporate temporal information in the analysis of movement data.An accurate description of space use is essential to investigate movement. The Brownian Bridge movement model is a formal description of animal space use. It integrates the estimated position of the animal between observed locations over time, assuming continuous random movement. I extended the model to account for changes in the amount of movement to describe heterogeneous trajectories (Chapter 3). Using this technique it is possible to describe trajectories containing various behaviours and life history stages. For example, tracks that contain both migration and breeding behaviours, or trajectories with active and non-active periods. The Bivariate Gaussian Bridges further generalize the Brownian Bridges, but separate movement variance in two components on in the direction to the next location and one perpendicular to this direction. I can show that the decomposition of the movement variance describe correlated random walks better and produce an equal or better fit for trajectories of various species (Chapter 4). These models provide a more accurate description of space use. This in turn will make it possible to investigate a multitude of questions with higher accuracy and precision.Based on this work, I quantify the space use of migrating raptors. From an accurate description of the individual space use an environmental model for migration can be calculated per species. The model is used to predict species-wide space use during migration between winter and summer ranges. This creates a global map of migration diversity and intensity across species based on observed migrations (Chapter 5).Migration routes are not only shaped by static environmental conditions. To explore how migratory routes could be shaped by dynamic environmental conditions I use two decades of wind data to calculate the quickest migration route for a continuously flying bird. These routes are specific to one starting time, because wind conditions change continuously,and require knowledge of future wind conditions to be encountered. This makes it virtually impossible for individual animals to predict these routes at the time of departure. However, provided spatial and temporal persistence evolutionary processes could theoretically select at population level for adaptive adjustment of migratory paths. I investigated whether some of these routes could be followed every year on the basis of selection for the shortest travel time. There is nearly always an alternative route that is quicker than the shortest route over the years. This shows, that besides optimizing the timing of migration, optimization of the route is also important, that can be done by following a static route despite changing wind conditions (Chapter 6).With thesis I contribute to the study of animal movement by developing and applying various analytical techniques for movement research. The emphasis of these methods is especially on including time in the analysis. The methods developed here, together with other methods are used to investigate bird migration. These studies show how movement can be investigated on a global scale.