PARENT SESSION
Oral Session #74: Spatial Ecology: Models and Methods. Presiding: C. Loehle.
Friday, August 10, 2001. 8:00 AM to 11:30 AM. Hall of Ideas E.

An evaluation of neutral and non-neutral landscape models using patch size distributions and semivariograms.

Oline, David¹, Grant, Michael¹, ¹

ABSTRACT- Following the introduction of neutral landscape models to test for the existence of non-random spatial patterns, a variety of "non-neutral" landscape models have been devised to more closely mimic the landscape patterns observed in natural environments. Although many different models may be tweaked to closely mimic a given observed pattern, we ask how well is a given model able to exhibit the full range of behavior of natural landscapes? We investigate four existing models (random percolation, hierarchical random, fractal, and modified random clusters) and develop two new cellular automata-based models (iterated averaging and stochastic weighting). To evaluate the models, we use four landscape metrics: 1) contagion index, 2) scaling exponent (the log-log slope of total variance as a function of pixel width), 3) semivariograms, and 4) patch size distributions (the probability of a patch being larger than an area A as a function of A, used to calculate the Korcak exponent). We establish the possible ranges of these metrics based on empirical data and theory, and then evaluate how much of the possible full range each model is able to exhibit by changing the model parameters. We suggest that those models which explicitly incorporate multiple scales and use cellular automata rules produce more of the full range of behavior, while models which begin with an initial random seeding of the landscape fare more poorly.

KEY WORDS: neutral landscape models, cellular automata, patch size distributions, semivariograms