Production of stream habitat gradients by montane watersheds: hypothesis tests based on spatially explicit path analyses

We studied how the features of mountain watersheds interact to cause gradie nts in three stream attributes: baseflow stream widths, total alkalinity, a nd stream slope. A priori hypotheses were developed before being tested in a series of path analyses using data from 90 stream reaches on 24 second- t o fourth-order streams across a fifth-order Rocky Mountain watershed. Becau se most of the conventional least squares regressions initially calculated for the path analyses had spatially correlated residuals (13 of 15 regressi ons), spatially explicit regressions were often used to derive more accurat e parameter estimates and significance tests. Our final working hypotheses accounted for most of the variation in baseflow stream width (73%), total a lkalinity (74%), and stream slope (78%) and provide systemic views of water shed function by depicting interactions that occur between geomorphology, l and surface features, and stream attributes. Stream gradients originated ma inly from the unidirectional changes in geomorphic features that occur over the lengths of streams. Land surface features were of secondary importance and, because they change less predictably relative to the stream, appear t o modify the rate at which stream gradients change.