DISENTANGLING INTERPOLATION AND EXTRAPOLATION UNCERTAINTIES IN SPECIES DISTRIBUTION MODELS: A NOVEL VISUALIZATION TECHNIQUE FOR THE SPATIAL VARIATION OF PREDICTOR VARIABLE COLINEARITY

摘要: Abstract. - Species distribution models (SDMs) are increasingly used in many scientific fields, with most studies requiring the application of SDM to predict likelihood occurrence and/or environmental suitability locations and time periods outside range data set fit model. Uncertainty quality predictions caused by errors interpolation extrapolation has been acknowledged for a long time, but explicit consideration magnitude such is, as yet, uncommon. Among other issues, spatial variation colinearity predictor variables development SDMs may cause misleading when applying novel periods. In this paper, we provide framework spatially identification areas prone changes inter-correlation structure (i.e. their colinearity) predictors development. The proposed method is compatible all algorithms currently employed, expands available toolbox assessing uncertainties raising from predictions. We an implementation analysis script R statistical platform online appendix.