ANALYSIS OF PATTERNS IN HIERARCHICALLY STRUCTURED LANDSCAPES

Scale-dependent effects of landscape heterogeneity were examined by a series of hierarchically structured maps generated by randomly selecting site suitability at successively finer scales. Examination of the patterns (i.c., the number and size of habitat clusters, the total amount of edge on the map) generated by hierarchical maps shows that the global fraction of suitable sites, P, does not fully explain observed changes in landscape properties. Although p(i) (the fraction of suitable sites within hierarchical level i) must be greater than the critical threshold for connectance (p(c). = 0.5928), it is possible for maps to be fully connected when P is much less than p(c). Clusters are larger on hierarchical maps than random maps when p(i) is greater then p(c), but smaller when p(i) is less than p(c). The grain size within hierarchical levels also affects the structure of the maps, with fine-grained units producing higher total number of clusters than random maps. Several lines of evidence indicate that the patterns of landscape heterogeneity are scale-dependent, making hierarchically structured maps a useful tool for studying landscape patterns at different scales. Because biological processes are sensitive to pattern at particular scales, it is necessary to develop a means to describe and predict changes in landscape pattern with changes in scale.