Representations of real and virtual landscapes are often abstracted and quantified as "landscape metrics" in landscape ecology, and while much is known about them after 20 years of use, some relations are as yet unexplored. The objective of this research is to gain a better understanding of the effects of spatial representation in landscape ecology. Nine hundred artificial landscapes were generated with varying spatial patterns created using a fractal landscape generator with 3D fractal dimensions ranging from 2.1 to 2.9 in increments of 0.1. The 3D landscapes were reduced to 2D patterns by putting the cells into five classes. Forty-four landscape metrics were used to measure the resulting 2D landscape spatial patterns, and the relations of nine of these to the generating 3D fractal dimensions and to each other across these fractal dimensions were analyzed. The expected nonlinear relations between metrics and the generating 3D fractal dimensions led to unimodal correlation structures between some of the metrics, with low, nonsignificant correlations in the middle range of the 3D fractal dimensions. The selection of metrics and their interpretation, especially in the midrange of landscape complexity, may need to be reassessed.
Available at: http://works.bepress.com/george_malanson/48/