Microbial colonization of artificial substrates introduced into aquatic ecosystems is affected by the relative levels of nutrients and toxicants. The producitivity of microbial biota integrates several factors affecting organism survival and is expressed in relative rates of propagule production and colonization. It is now possible to examine factors affecting microbial colonization in laboratory microcosms. Colonized artificial substrate species sources can be used in the laboratory to measure nutrient and toxicant effects. Studies have shown that microbial colonization is at least as sensitive a technique as long-term single-species testing and has allowed controlled measurement of complex community responses to disturbance. It is possible to model the nonlinear colonization process and to compare colonization rates, equilibrium species numbers, and biomass production in test systems. Such testing can examine effects of nutrients, pure toxicants, or complex effluents. The authors found adverse effects on species dispersal in static test systems for cadmium and copper at concentrations of <1 ug Cd/L and 18 ug Cu/L, although low levels of copper enhanced the species numbers. In microcosm systems receiving continuous toxic input, we found adverse effects of chlorine on species dispersal at 2 ug/L; however, net production was elevated at concentrations up to 100 ug/L. In tests with the herbicide atrazine, the number of species and the net production were stimulated at low levels (3 to 30 ug/L). Stimulation of the species number and production may be a result of effects on control mechanisms and not the result of a subsidy to the community. The study of natural community dynamics in evaluating the effects of toxic materials provides evidence of effects on the emergent properties of systems, which are not available from studies of individual species. In many cases, effects on communities may occur at concentrations similar to those producing effects on sensitive single species. However, testing at the community or system level can provide predictions of community or system components most at risk. As such, system level tests should provide predictions that can be directly validated in ecosytems receiving toxic inputs.