Autocorrelation in Logistic Regression Modelling of Species' Distributions

Abstract

Logistic regression models are becoming widely used for modelling plant and animal distributions, but most studies ignore the spatial arrangement of the data and its effect on the models. One method of including this spatial information is presented, with reference to the distribution of Oxyria digyna (L.) Hill (mountain sorrel) in Britain. The advantages and disadvantages of explicitly modelling the autocorrelation in the data are discussed.