ABSTRACT
Well-established pattern recognition and classification modeling techniques ap plied to large, multivariate databases generated from national-or regional-scale environmental data synthesis programs can be useful in addressing environmental problems. For example, these methods may be combined with existing geochemi cal modeling techniques to identify, by means of readily available major ion chemical data, areas in which the concentrations of selenium in surface water may be harmful to waterfowl and fish. Classification models based on these regional-or national-scale data synthesis programs can then be used by regulatory and management personnel to assess selenium toxicity potential in areas for which data are available for major ion concentrations but not for selenium concen trations.
