Spatial-temporal models for counting processes in ecology

Spatial and temporal counting of individuals is often necessary in ecological population study. Meanwhile, independence between counts is generally assumed for the statistical data analysis although the individuals behavior or the sampling plan do not always justify such an assumption. In this paper, spatio-temporal models for individuals distributions, taking into account the proximity between sampling units, are presented. The case of negative binomial conditional or marginal laws is mainly discussed. It is emphasized that more care must be taken in using certain statistical methods because situations arise for which they are not appropriate.