Hierarchical Bayesian methods estimate invasive weed impacts at pertinent spatial scales

Invasive weed impact estimates are needed to determine whether or not weeds warrant costly control measures. Typically, land managers seek local weed impact estimates (e.g. ranches, parks) and policy-makers want to know how weeds are impacting entire regions. Our goal was to provide local and regional impact estimates for a ubiquitous invasive weed: leafy spurge (Euphorbia esula L.). The specific impacts we looked at related to desired species biomass production, livestock carrying capacities, and grazing land values. Our basic approach was to use an empirical model that characterizes weed biomass across the landscape in combination with another empirical model that predicts weed impact from weed biomass. Our investigation revealed that, without on-site plant biomass data, site-specific leafy spurge impacts are highly uncertain. Supplementing our general predictive model with small quantities of on-site data increased precision considerably. For the 17-state region we considered, 95% Bayesian credibility intervals indicated leafy spurge reduces cattle carrying capacities by 50–217 thousand animals a year and reduces grazing land values by 8–34 million dollars a year. Additional plant biomass data from randomly selected, leafy spurge-infested sites would shrink these fairly wide intervals.