Diatom-based stream bioassessment: The roles of rare taxa and live/dead ratio

It is unclear whether accounting for the number of rare taxa or differentiating live (cells with visible chloroplasts) and dead (empty cells) diatoms would enhance the accuracy and precision of diatom-based stream bioassessment. My dissertation research examines whether the number of rare taxa and percentage (%) live diatoms can be used as indicators of human disturbance in streams/rivers. To address my objectives, I analyzed two datasets collected at different spatial scales. The large scale dataset was collected over the course of five years and included more than 1300 sites and over 200 environmental variables from 12 Western US states (US EPA's Environmental Monitoring and Assessment Program). The small scale dataset included 25 streams from the Oregon Coast Range. My findings revealed that rarity definition is important in bioassessment. Rural taxa richness (number of taxa with high occurrence and low abundance) was the only rarity metric that distinguished least disturbed (reference) sites from the most disturbed (impacted) ones. Neither of the other two rarity metrics, satellite (taxa with low occurrence and abundance) and urban (taxa with low occurrence and high abundance) taxa richness, was able to do that. Results from regression tree analysis revealed that rural taxa richness increased with human disturbance, but it was significantly higher at impacted sites only in the Mountains ecoregion (t-test, p<0.05). Neither of the other two ecoregions (Plains and Xeric) exhibited significant differences between reference and impacted sites (t-test, p>0.05). Percentage live diatoms distinguished reference from impacted sites only in the Mountains ecoregion (t-test, p=0.02) and somewhat in the Plains (t-test, p=0.05). However, % live diatoms exhibited opposite patterns in the two ecoregions. They increased with disturbance in the Mountains and decreased in the Plains. The results from the small scale study in the Oregon Coast Range revealed similar species compositions between live and live+dead diatom assemblages (non-metric multidimensional scaling) and similar relationships with their environmental variables (linear fitting). Both assemblages correlated well with in-stream physical habitat conditions (e.g., channel dimensions, substrate types, and canopy cover). Both rural taxa richness and % live diatoms can be used as indicators of human disturbance in streams/rivers, especially in the Mountains ecoregion, dependent on specific assessment objectives.