SPATIAL HETEROGENEITY AGAINST HETEROSCEDASTICITY - AN ECOLOGICAL PARADIGM VERSUS A STATISTICAL CONCEPT

One widely spread ecological perspective, at least in landscape ecology, is based on the dictionary definition of "heterogeneous': consisting of dissimilar or diverse ingredients or constituents, mixed. From a statistical viewpoint, heterogeneity, when applied to the distribution of the values taken by a random variable, is the opposite of homogeneity, which refers to sameness or similarity. In a spatial context, the acceptation given to the word "heterogeneity' primarily depends on the type of spatial pattern of points and their distribution in space, while for a surface pattern, it concerns instead the values taken by a qualitative or quantitative variable. The ecological and statistical aspects of spatial heterogeneity are compared; in particular, discussion from an ecological viewpoint focuses on fundamental aspects of spatial heterogeneity and its functional role in ecosystems, combined or not with temporal variability. While heterogeneity is a term commonly used by the ecologists and statisticians, heteroscedasticity is a purely statistical concept which concerns a particular type of heterogeneity restricted to inequality of variances. Difficulties generated by heteroscedasticity in statistical data analyses and solutions taking it into account are reviewed. -from Authors