If certain ecological traits convey adaptive advantages to taxa that possess them, the frequency of taxa within assemblages possessing these traits should vary in a predictable manner as environmental conditions change. We tested this idea by using data from 729 reference-quality stream reaches in the western USA to model the expected probabilities of capture of 437 taxa as a function of natural environmental setting. We assigned both functional traits (sensu Poff et al 2006) and tolerance values (sensu Carlisle et al 2007) to most of these taxa. We then determined which taxa increased or decreased in frequency of detection under agriculture (n=80), urban (n=85), and hydrologically modified (n=35) settings and if the presence of ecological traits could predict the direction of response by 135 different increaser or decreaser taxa. Tolerance values were slightly more successful in predicting taxa responses (80% correct) than functional traits (75%), perhaps because tolerance values were more finely resolved (10 increments) than functional trait states (0/1). We could not predict the type of landscape/waterway alteration affecting a stream with either type of trait, which implies that most types of ecosystem alteration may create similar forms of stress.