Dispersal across continuous and binary representations of landscapes

Many simulations of plant or animal dispersal across landscapes fragmented by human activity have represented the environment as habitat or non-habitat. Here, the consequences of representing habitat on grids as a continuous surface (0-1) versus as binary (0, 1) are examined for such models. Simulated landscapes with mean habitat quality of 0.5, with different variance, and with 50% each of habitat and non-habitat are compared. Random and non-random patterns are simulated. Dispersal can be directionally biased. Dispersal across continuous landscapes is more frequent and faster than across binary landscapes. On continuous landscapes, pattern matters, with success and speed reduced as local heterogeneity increases. Increased variance also reduces success and speed. The degree to which a species uses the landscape as a continuum of habitat, versus as either/or, will improve its response to forces that require increased movement for survival, such as habitat fragmentation and climatic change. The range of representations from binary to continuous will affect models including spread of disturbances, but the importance of heterogeneity will vary with the species or phenomena being modeled. (C) 2003 Elsevier B.V. All rights reserved.