Alterations to water quality and quantity elicit similar stream fish functional trait responses in three North American rivers

• Ecological indicators based on functional traits link fishes to their environment. • This study further explores fish traits in relation to differing anthropogenic alterations. • Most traits responded to dewatering, followed by flow alteration and water pollution. • We found three indicator traits linked to habitat, life history, and trophic strategy. • Trait response direction was consistent across three rivers differing in alteration. Growing human population and demand for freshwater have led to global alterations to rivers, including construction of millions of reservoirs that alter flow regimes, groundwater depletion that causes channel dewatering, and nutrient pollution that degrades water quality. Although taxonomic-based indicators have played a central role in measuring ecological integrity of rivers, trait-based ecological indicators may be applied across regions and provide insight into the mechanistic linkages between environmental variation and organism responses. The goal of this study was to conduct a multi-system assessment of fish functional trait responses to anthropogenic alteration of rivers. Our interests were to determine if (1) traits commonly available for freshwater fishes responded through time as rivers transitioned across states of alteration, and (2) the number of freshwater fish strategists for each trait responded in the same direction across rivers afflicted primarily by dewatering, flow regulation, or water quality improvement. We compiled fish community data from three case studies and used existing trait databases to calculate trait strategist richness defined as the number of species expressing a trait. We then fit random forest models and used recursive feature elimination to identify the traits that were most important for predicting the period from which replicated fish collections were taken, thereby providing a test of which traits most strongly correlated with environmental alterations. Finally, we use partial dependence plots to assess if traits responded in the same directions across case studies and therefore might represent broadly applicable ecological indicators. The largest number of traits responded to stream dewatering (n = 16), with fewer traits responding to water quality abatement (n = 5) and flow homogenization (n = 4). Across case studies, fishes classified as partial herbivores (i.e., they consumed plant matter in addition to other items) and fishes classified as intermediate between periodic and opportunistic life history strategy endpoints declined as flows were depleted, homogenized, or polluted. Fishes using gravel substrates declined as flows were depleted and polluted, but increased after flow homogenization caused by dam construction, perhaps because of altered sediment dynamics. Although the magnitudes and thresholds for change in strategist richness we identify here are specific to the rivers we studied, the emergent themes of consistency in responsiveness and directionality suggest (1) the strategists richness metric and (2) traits related to trophic ecology, life history, and habitat use might be useful indicators for detecting and measuring ecological change in other river systems.