they are broken down by sunlight over a period of decades to several centuries. CFCs released in the 1990s will continue to destroy ozone in the stratosphere well into the twenty-first century and, in some instances, beyond. Because of such slow effects and the interdependencies of environmental systems, many human interventions in the global environment constitute uncontrolled experiments whose results may not be known for generations. This makes knowledge difficult to accumulate. It also increases the demand for knowledge both because these experiments may threaten the whole earth and because they have the potential to set catastrophes in motion before their effects are even noticed.
Global environmental changes can result from the interactions of local systems with each other and with larger-scale systems. For some analytic purposes, it is inadequate to treat the earth as possessing a single environment. Although the atmosphere is global, understanding of the biosphere may need to be built up from knowledge at smaller spatial scales, such as ecosystems or biomes. Thus, knowledge of global change requires ways to understand relationships across spatial scales (Clark, 1987; Rosswall et al., 1988). Human activities compound the challenge by redistributing species and transforming habitats, thus altering the ways ecosystems interact.
These characteristics of the global environment present serious challenges for scientific research and may call for new theories and methods. In addition to progress on scientific questions that fall within standard disciplinary boundaries, problems of global change require approaches that treat the earth as a single interactive system and stress the powerful interdependencies among environmental (and human) systems. Such approaches tend to be interdisciplinary rather than multidisciplinary (Schneider, 1988) and are often characterized by holistic analytic premises such as those of ecology or systems analysis.
The nature of the global environment also raises doubts about the value of the existing structure of scientific disciplines for understanding global change. To the extent that resources continue to be channeled through the familiar disciplines, the disciplines look increasingly like part of the problem. Those working on computer models of global climate change (that is, general circulation models) already find it necessary to incorporate into their algorithms variables and relationships from various disciplines of physical science; biological variables and relationships will increasingly be included as the models are refined. And pressure is growing to incorporate projections of human activities into