Can diatom motility indices reflect excess fine sediment condition in streams?

The diatom motility trait is widely used in bioassessment studies, however, there no strong consensus as to what environmental conditions reflects. We random forest models explore the behavior of several diatom-based indices and examined whether stronger motility-environmental relationships could be developed by controlling for factors that influence natural sediment loads co-occur with anthropogenic loading. Across study area, median values most stressors were low; streambed fine sediments ranged from 0 100% (median: 6.7%) total phosphorus concentrations 2587 µg L?1 17 L?1). Depending on index, abundance motile taxa between 8 47%, was highest plains ecoregion. Few sites dominated taxa, only 9 having a relative highly >50%. Refinement did result improved model performance; RF had lowest RMSE (0.11) variables explaining 34.9% variability index. Variables indicative water column not important predictors any index refinement improve relationship sediments. related quality (e.g., phosphorus, conductivity), flow runoff, % channel fast moving water), watershed characteristics slope) all indices, while lower importance values. Stratifying ecoregion resulted slight improvement performance mountain (RMSE = 0.09) but diatom-motility instream also improvements 0.10), conditions. geomorphic features effect how responded Our results show complex. Strong may difficult detect large-scale observational studies covering complex landscapes multiple interacting factors.