Composition and modeling of riparian vegetation in the West Fork of the Gallatin River watershed
Loading...
Date
2007
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Montana State University - Bozeman, College of Agriculture
Abstract
Riparian areas contribute to the health of watersheds through their influence on hydrologic, biogeochemical, physical, and ecological processes. Limited research has focused on riparian systems of small mountain watersheds in the western U.S., which are increasingly under pressure from development activities. Watershed managers would benefit from an increased understanding of environment-riparian relationships in mountainous watersheds, for the purpose of assessing habitat and potential available nutrient buffering. This study assessed vegetation-environment relationships using digitally-derived terrain variables and wetland indicator scores, and used these relationships to assess the composition and model the cross-valley extent of riparian vegetation within the West Fork of the Gallatin River watershed in southwest Montana. Digital terrain analysis was used to extract the following terrain predictors: elevation above and distance from streams, plot gradient, valley width, and a topographic wetness index, which integrates the upslope area that contributes flow to a plot, and the plot gradient, thus serving as a measure of site wetness. Species abundance was used to assign weighted plot wetland indicator scores in order to focus on cross-valley gradients, with plots below a threshold score (mesic plots) designated as riparian plots.
Linear regression, and two types of generalized regression (additive and linear) were used to assess relationships between terrain predictors and (a) individual species, (b) species associations based on clustering of plots, (c) plot wetland indicator scores and (d) riparian plots as defined by the threshold wetland indicator score. Elevation above the stream and plot gradient were the strongest predictors, followed by the topographic wetness index. While generalized additive models often resulted in higher D2 values (e.g. r2), they were often overfit when using plot wetland indicator scores as a response variable. Therefore generalized linear models were the strongest models for predicting the cross-valley extent of riparian vegetation. Lastly, prediction of the shrub strata produced stronger models than the herb strata. This study demonstrated that digitally-derived terrain variables and wetland indicator scores provide a useful method of assessing riparian vegetation-environment relationships, which can be used to aid researchers and managers in better understanding riparian systems within mountainous watersheds of this region.
Linear regression, and two types of generalized regression (additive and linear) were used to assess relationships between terrain predictors and (a) individual species, (b) species associations based on clustering of plots, (c) plot wetland indicator scores and (d) riparian plots as defined by the threshold wetland indicator score. Elevation above the stream and plot gradient were the strongest predictors, followed by the topographic wetness index. While generalized additive models often resulted in higher D2 values (e.g. r2), they were often overfit when using plot wetland indicator scores as a response variable. Therefore generalized linear models were the strongest models for predicting the cross-valley extent of riparian vegetation. Lastly, prediction of the shrub strata produced stronger models than the herb strata. This study demonstrated that digitally-derived terrain variables and wetland indicator scores provide a useful method of assessing riparian vegetation-environment relationships, which can be used to aid researchers and managers in better understanding riparian systems within mountainous watersheds of this region.