Citation:

McManus, M., E. DAmico, E. Smith, R. Polinsky, J. Ackerman, AND K. Tyler. Spatial Predictions in Rivers and Streams: Conductivity in an Appalachian Stream Network. The Nature Conservancy Lunch and Learn Series, N/A, Virtual, April 29, 2022.

Impact/Purpose:

Freshwater rivers and streams are imperiled by multiple stressors, such as secondary salinization, the increase of anthropogenically derived salts in freshwater, and elevated nutrient concentrations, which are associated with eutrophication.  Addressing these threats may require acknowledging the spatial network structure of rivers and streams so we can predict where these stressors occur.  We predicted specific conductivity in streams throughout the 525 stream kilometers of an Appalachian watershed.  We used water quality monitoring data with land cover geographic information systems (GIS) data in a spatial stream network model that included the spatial autocorrelation from 60 monitoring sites in the Right Fork Beaver Creek watershed in Floyd and Knott Counties of Eastern Kentucky.  We showed that the correlations in conductivity between monitoring sites differed under high versus low discharge conditions of the watershed.  By being able to predict water quality, such as specific conductivity, we can better understand how land cover and use in the watershed impacts the stream biota.  We also describe ongoing work to combine different water quality monitoring data to make predictions of river nutrient concentrations.

Description:

Freshwater rivers and streams are imperiled by multiple stressors, such as secondary salinization, the increase of anthropogenically derived salts in freshwater, and elevated nutrient concentrations, which are associated with eutrophication.  Our goal is to understand how stream site measurements on the network can let us assess and predict stressors over an entire watershed. Watersheds are dynamic, with stream networks expanding and contracting, creating differences in connectivity along longitudinal, lateral, and vertical dimensions. In this study, the stream network was repeatedly sampled under very different discharge conditions. What I want to know is how do such different conditions affect spatial predictions of stream conductivity?  Conductivity is an aggregate measure of total dissolved ions, which can affect stream biota.  I use spatial stream network models to predict stream conductivity and show that incorporating spatial autocorrelation among the monitoring sites improves those predictions.  The different discharge conditions altered the spatial dependency of conductivity among the sites.  I will also describe ongoing research to make statewide predictions on stream nutrient concentrations by pooling different types of water quality monitoring data.  By judiciously combining these data sets we can achieve the sample size, spatial configuration of headwaters to watershed outlets, and spatial density of sites that allow us to go from condition assessments to statewide predictions of river and stream nutrient concentrations in watersheds.

Record Details: