Effectiveness of RIVPACS predictive models to evaluate diatom response to nutrient stress in coastal California streams
Thesis (M.S.) Division of Science and Environmental Policy
The goal of this project was to determine if predictive models of diatom assemblages would provide an effective method to report on biological degradation in streams along the Central Coast of California. This project focused on nutrient stress to evaluate stream water-quality degradation. I employed the River Invertebrate Prediction and Classification System (RIVPACS) model with diatom assemblages. Diatoms were an accessible indicator of nutrient stress occurring in abundance on Central Coast streams. Diatom samples from 190 stream sites were used to construct and test the RIVPACS model. The RIVPACS methodology used a reference condition approach to compare assemblages at reference sites to observed assemblages at degraded test sites. A ratio of observed taxa to expected taxa (OE) was the concluding measure of biological integrity at each site. I used the OE scores to test the postulate that degraded sites had diatom assemblages dissimilar from the reference site diatom assemblages. The RIVPACS model did not performed well. The model suffered from low precision of reference site OE scores (mean SD = 0.22) and lack of accuracy to consistently predict low OE scores at known degraded sites. However, the model was able to identify likely trends. The uncertainty in the RIVPACS model did not provide a definitive measure of model effectiveness. I concluded the assessment model was limited by the quality of reference streams and the temporal variability and spatial patchiness of diatom assemblages.