THE AGGREGATION OF IMPACTS: USING SPECIES-SPECIFIC EFFECTS TO INFER COMMUNITY-LEVEL DISTURBANCES.

This article discusses the appropriate role for natural field experiments in ecology. At one extreme is the argument that experiments are always the preferred way to carry out ecological research, while at the other extreme is the assertion that classical experimentation in ecology is essentially impossible. As with most such debates the truth is probably somewhere in the middle, and even those who champion the experimental approach would concede that the application of standard experimental-design principles to ecology is often exceedingly difficult. For example, due to the scale at which ecosystem processes operate, true replication rarely exists. Furthermore, because many of the factors affecting ecosystems are non-manipulable, the standard strategy of randomization to equalize treatment and control groups is generally not available. These same difficulties also occur in the design of ecological impact studies, but there are additional complications. In natural experiments the assignment of an impact to a site is typically not under the control of the researcher and the simultaneous imposition of impacts on suitable replicate sites rarely occurs.