A subwatershed base map of 84 hydrologic subregions within the Columbia River Basin (approximately 58,361,000 ha) was developed following hierarchical principles of ecological unit mapping. Our primary objectives were to inspect the relations between direct and indirect biophysical variables in the prediction of valley bottom and stream type patterns, and to identify hydrologic subregions (based on these results) that had similar aquatic patterns for which consistent management practices could be applied. Realization of these objectives required: (1) stratified subsampling of valley bottom and stream type composition within selected subwatersheds, (2) identification of direct and indirect biophysical variables that were mappable across the basin and that exerted primary control on the distribution of sampled aquatic patterns, and (3) development of hydrologic subregion maps based on the primary biophysical variables identified. Canonical correspondence analysis indicated that a core set of 15 direct variables (e.g., average watershed slope, drainage density, ten-year peak flow) and 19 indirect variables (i.e., nine subsection groups, four lithology groups, and six potential vegetation settings) accounted for 31 and 30 percent (respectively) of valley bottom/stream type composition variability and 84 and 80 percent (respectively) of valley bottom/stream type environmental variability within subsamples. The 19 indirect biophysical variables identified were used to produce an ecological unit classification of 7,462 subwatersheds within the basin by a hierarchical agglomerative clustering technique (i.e,, hydrologic subregions were identified). Discriminant analysis indicated that 13 direct biophysical variables could correctly assign 80 percent of the subwatersheds to their indirect biophysical classification, thus demonstrating the strong relation that exists between indirect biophysical based classifications (ecological units) and the direct biophysical variables that determine finer-level aquatic patterns. Our hydrologic subregion classifications were also effective in explaining observed differences in management hazard ratings across all subwatersheds of the basin. Results of this research indicate that ecological units can be effectively used to produce watershed classifications that integrate the effects of direct biophysical variables on finer-level aquatic patterns, and predict opportunities and limitations for management.
