Modelling in landscape ecology – regionalisation by means of habitat modelling

Figure 1: Principle of habitat modelling: sampling of species distribution data (here: presence-absence data) and selected predictor variables; estimating an empirical, predictive model (in this case: by logistic regression a generalised linear model for binary response variables, other, more flexible approaches are generalised additive models, classification and regression trees, artificial neural networks) and model evaluation; analysis of relevant driving variables and habitat factors; prediction and regionalisation of predicted occurrence probabilities if predictors are available as maps; external and/or internal model validation by either model transfer or resampling techniques. Modelling in landscape ecology – regionalisation by means of habitat modelling