Streambank erosion is an important contributor to sediment and nutrient export. This study determined total P (TP), soil-test P (Modified Morgan's, MM-P), and the degree of P saturation (DPS) in eroding riparian soils along four Lake Champlain Basin stream corridors. We investigated the relationship between these data and soil texture and with a series of GIS-derived landscape metrics. We also quantified the potential P load from eroding streambanks using remote sensing. Soil samples were taken from 76 erosion features to a depth of 90 cm on four streams in Chittenden County, Vermont. Mean concentrations of TP and MM-P were similar among the watersheds and through depth. Neither TP, MM-P, nor DPS were well related to texture. Metrics from available spatial databases for parent material, soil series, and landscape position were somewhat useful in predicting TP and MM-P. Eroding streambank soil from 2004 through 2007 in the four streams was estimated to contain from 0.5 to 3.9 Mg of TP and 1.4 to 10.9 kg MM-P. The mean DPS in each watershed was <18% and, along with low MM-P concentrations, suggests that eroded streambanks may act as sinks rather than sources of P. The portion of total nonpoint P export potentially contributed by streambank erosion ranged from 6% in the stream with the lowest erosion rate to 30% in the stream with the highest erosion rate. Based on TP values, the P contribution of these streambanks could be considerable, and more information is needed on their actual contribution to bioavailable P in receiving waters.

Roads in rural, upland landscapes are important sources of runoff and sediment to waterways. The downstream effects of these sources should be related to the connectivity of roads to receiving waters. Recent studies have explored this idea, but only simple metrics have been used to characterize connectivity and few studies have quantified the downstream effects of road–stream connectivity on sediment or solute budgets and channel morphology. In this study, we evaluated traditional and newly developed connectivity metrics that utilized features of landscape position and delivery pathway to characterize road–stream connectivity in upland settings. Using data on stream geomorphic conditions developed by the Vermont Agency of Natural Resources (Montpelier, VT), we related road connectivity metrics to channel condition on a set of 101 forested, upland streams with minimal development other than predominantly gravel road networks. Logistic regression indicated that measures of road density, proximity and orientation successfully distinguished among categories of stream geomorphic condition at multiple geographic scales. Discriminant function analysis using a set of inherent channel characteristics combined with road connectivity metrics derived at the reach corridor scale successfully distinguished channel condition for over 70% of the channels evaluated. This research contributes to efforts to evaluate the cumulative downstream effects of roads on stream channels and aquatic resources and provides a new means of watershed assessment to derive metrics that can be used to predict channel condition. Copyright © 2014 John Wiley & Sons, Ltd.