Factors Controlling Structural and Floristic Variation of Riparian Zones in a Mountainous Landscape of the Western United States

We examined landscape patterns in the physical conditions and vegetative composition of montane riparian zones to identify their most important sources of variation. Information on plant species cover and on physical characteristics that occur at coarse, medium, and fine scales was collected for 144 riparian plots located throughout the Lake Tahoe Basin, which straddles the California-Nevada border in the western United States. Constrained and unconstrained ordination analyses were used to identify the most important correlates of physical form and plant species composition. Through multivariate analysis of environmental variables (principal components analysis), vegetation data (detrended correspondence analysis), and the combined relationship between the environmental and vegetation data (canonical correspondence analysis), we consistently found that the greatest variation occurred along a gradient of decreasing valley width, decreasing stream sinuosity, and increasing stream slope. Although surface characteristics reflected a 2nd important source of variation in physical conditions, plant species distribution was not strongly correlated with riparian surface conditions. Strong correlations among physical variables that occur at different scales, such as between valley form and geofluvial surface and between geofluvial surface and surface conditions, support the use of a physically based hierarchical framework for organizing riparian zones within the landscape. Such a hierarchical framework would be useful for interpreting patterns in riparian structure and process at different scales and could be applied to riparian zones in other mountain landscapes of the western United States and elsewhere. Moreover, our finding that riparian plant species composition is most strongly correlated with environmental variables that occur at coarse to moderate scales, most of which can be derived from existing data, supports the idea that modeling montane riparian community distribution using topographic and remotely sensed data could be useful; however, a large degree of species variation, unexplained by the variables we collected, indicates that other variables, perhaps disturbance regime, should be included in such a venture