Figure 2. Combined factor and path analysis for B. truncatus biomass demonstrating top-down and bottom-up effects of insecticides and fertilizer, respectively. Top-down effects: insecticide exposure reduced final crayfish counts, which directly increased snail biomass by reducing predation, and also indirectly increased snail biomass by increasing snail habitat H. verticillata (crayfish ate H. verticillata ). Bottom-up effects: Fertilizer increased periphyton and H. verticillata , which increased snail biomass because they are food and habitat, respectively, for snails. Herbicides reduced H. verticillata , snail habitat, and, thus, on average, had net negative effects on snails. P-values for paths in the model are reported below each standardized coefficient. Red arrows denote significant negative paths; black arrows denote significant positive paths; dashed arrows denote non-significant pathways. Boxes represent directly measured variables and circles represent latent variables. Indicator variables for latent variables have been omitted from the figure to reduce visual complexity, but are reported in Table S5. The path model had a good fit to observed data (CFI = 0.99, χ2 = 26.3, df = 26,p-value = 0.45).