Trophic cascades are extensively documented in nature, but they are also known to vary widely in strength and frequency across ecosystems. Therefore, much effort has gone into understanding which ecological factors generate variation in cascade strength. To identify which factors covary with the strength of cascades in streams, we performed a concurrent experiment across 17 streams throughout the Sierra Nevada Mountains. We eliminated top consumers from experimental substrates using electrical exclusions and compared the strength of indirect effects of consumers on the biomass of primary producers relative to control patches. In each stream we 1) classified the dominant invertebrate herbivores according to life-history traits that influence their susceptibility to predators, 2) determined the abundance and diversity of algae and herbivores, and 3) measured production-to-biomass ratios (P:B) of the stream biofilm. This allowed us to assess three common predictions about factors thought to influence the strength of trophic cascades: cascade strength 1) is weaker in systems dominated by herbivores with greater ability to evade or defend against predators, 2) is stronger in systems characterized by low species diversity, and 3) increases with increasing producer P:B.
When averaged across all streams, the indirect effect of predators increased the biomass of periphyton by a mean 60%. However, impacts of predators on algae varied widely, ranging from effects that exacerbated algal loss to herbivores, to strong cascades that increased algal biomass by 4.35 times. Cascade strength was not related to herbivore traits or species diversity, but decreased significantly with increasing algal diversity and biofilm P:B in a stream. Partial regression analyses suggested that the relationship between cascade strength and algal diversity was spurious, and that the only significant covariate after statistically controlling for cross-correlations was algal P:B. Our study contributes to the ongoing debate about why trophic cascade strength varies in nature and is useful because it eliminates factors that have no potential to explain variation in cascades within these stream ecosystems.