• INSTAR News
• Instar Partners

• The Science
• GIS & Data Management
• Biological Sampling
• Fish Collection
• Macroinvertebrate Collection
• Habitat Assessment
• Assessment and Characterization
• Watershed Assessment
• Stream Reach Assessment
• Stream Model Development

• Downloadable Documents
• Glossary
• INSTAR Blog
• Tutorials

Stream Reach Model Development

Typically in stream assessments, a stream collection is compared to a reference condition to determine the extent of degradation or health. Based upon literature, choosing a nearby reference condition can be very challenging. From the graphic on the right, assessing a site based upon results obtained at a similar and close proximity location can be useful, however landscape alterations may be widespread throughout the watershed and may inhibit finding an appropriate reference stream/condition.

Empirical data and derived metrics are compiled and analyzed with multivariate techniques (e.g. correspondence analysis (CA), detrended correspondence analysis (DCA), canonical correspondence analysis (CCA), principal components analysis (PCA), and multiple regression). The site scores (i.e., coefficients from the final response model) are entered as the response variable and significant (P<0.05) biotic and abiotic variables and metrics are entered as explanatory variables. Finally, a series of reference stream models (i.e., virtual reference streams) are created for each ecoregion and stream order. We use Gower’s similarity index to compare empirical scores obtained from sampled stream sites and reaches to the appropriate regional reference stream, generating an index of stream health (i.e., Virtual Stream Assessment, VSA, score; range 0-100%) as a measure of percent comparability to the appropriate (virtual) reference condition model. The distribution of VSA scores for a random subset of stream reaches is used to generate breakpoints (mean ± 1 SD) for stream health categories displayed on the INSTAR interface.

Click on the image to see a distribution of scores

Current reference stream models for upper and lower Coastal Plain, lower Piedmont, eastern shore, and Shenandoah basin regions include variables representing fish and macroinvertebrate assemblage structure, instream habitat, and geomorphology, and have substantial explanatory power (R2 up to 0.74). This integrative approach eliminates many of the limitations typically associated with traditional bioassessment methods (e.g. RBP, IBI), including lack of appropriate reference sites and stream classifications that are based on a single ecological component (e.g. biotic versus abiotic, fishes versus macroinvertebrates) that may not be diagnostic under many conditions.