Trait-based analyses are promoted as a way to better understand the forces that drive patterns of adaptive radiation in evolutionary time and influence the assembly of communities in ecological time. Another potential advantage of trait-based analyses is that they are potentially more transferable across regions, which typically differ in species composition, than taxonomic-based analyses. We used a western-wide multivariate species distribution model derived from 659 reference-quality sites to estimate the predicted probabilities of capture (pc) of 267 taxa at both reference and 350 probabilistically selected sites sampled between 2018 and 2022 throughout the western USA. We then estimated both expected (Fo) and observed (Fe) frequencies of occurrence for each taxon as Fe = Sum (pc) and Fo = Sum (presences) for non-reference and reference sites, respectively. We also assigned 86 biological trait states to each taxon to assess which trait states were most strongly associated with whether taxa were increasers or decreasers. Dominant environmental stressors in western USA streams include alterations associated with dams, urbanization, agriculture, and livestock grazing. We found that 98 taxa were strong decreasers (Fo/Fe < 0.8) and 91 taxa were strong increasers (Fo/Fe > 1.2). Fo of common taxa decreased more often than that of rare taxa, which tended to increase in Fo more often than common taxa. However, only 5 trait states (slow development, rare occurrence in drift, synchronous emergence, short lifespan, and free-range attachment) were associated with differences in taxa composition between reference and non-reference sites. Despite 21-27% of these streams having lost > 28% of their local taxa, these trait states were only weakly correlated with differences in taxa composition between reference and non-reference sites (Mantel r values = 0.30-0.39). It was also difficult to interpret how shifts in these trait states were mechanistically related to either natural environmental gradients or variation in land use. Ecology has struggled as a discipline to develop strong, predictive theories. Moreover, the ecological literature is full of poorly designed studies or studies whose conclusions suffer from confirmation bias. True progress will come only when we become more rigorous in designing, interpreting, and evaluating the research we conduct.