On uncertainty in species distribution modelling

Abstract

Thesis investigates the impact of positional uncertainty in species data and explores if examining (global) spatial autocorrelation in predictors can be a solution to understand the models’ robustness to these kind of uncertainty; using local indicators of spatial association to investigate where the positional uncertainty affects the models more; addresses the problem of modelling (local) spatial association and introduces a new method to measure local spatial associations; explores and visualizes model uncertainty in species distribution modelling; provides a synthesis based on the previous chapters, and suggestions for the future works.