Landscapes are diverse in the frequencies and distributions of habitat types. Does landscape heterogeneity, as a higher order (i.e. at a pheno- menological scale coarser than that of organisms) of diversity, allow greater connectivity; does higher order diversity change higher order functions? Climatic change and habitat fragmentation require that plant and animal species move across varied landscapes. Mediterranean-type ecosystems provide contrasts in landscape heterogeneity. In order to assess potential impacts of either and thus to propose remedial measures, it is necessary to understand general classes of connectivity in landscapes. We developed six rules for movement (random, persistent, and directionally biased for both movement throughout habitat and restricted to habitat edges), which we simulated on three general landscape patterns (random, hierarch- ical, and fractal) with different degrees of habitat fragmentation. Organisms that move only on the edge of habitat patches are more successful with a moderate amount of fragmentation but are less successful overall and especially in the non-random landscapes. Particular structures can thus differentiate connectivity for different vectors. We propose that general rules derived from simulations can be applied to the identification of landscapes that are near critical levels of fragmentation and which will respond sensitively to small increases in fragmentation or small efforts at restoration. Mediterranean-type ecosystems are well suited to test this proposal because of contrasting patterns and changes in fragmentation.