Fig. 2: Latent interactions derived from cross-guild/stage sensitivity analysis (guilds were used in this specific study for model transferability).
The relevant vital rate (here, mortality due to flood events; scaled from 0 to 1) was changed by increments of 0.01 in the vicinity of its actual value and population size was recorded. At each increment, this was repeated independently up to 1000 times to achieve stable convergence on mean values. Sensitivity was calculated as the slope of the average population size of a guild regressed against the vital rate, which is analogous to taking the partial derivative of population size with respect to the vital rate of interest (Lytle & Merritt 2004). Only sensitivities with R2 > 0.3 were retained for network analysis. Because each guild contains six stages, the matrix depicts both within-guild and among-guild effects. The diagonal depicts self-effects, where an increase in mortality causes a decrease in abundance of that guild/stage. Positive relationships suggest competition, where an increase in mortality rate in one guild/stage allows another one to increase. HT, XS, HS, MM, and DS represent riparian plant guilds that possess similar vital rates (respectively, hydroriparian tree, xerophytic shrub, hydroriparian shrub, mesic meadow, and desert shrub) and numbers indicate stage class.
