A comparison of deterministic and stochastic models for predicting the impacts of different sheep body lice (Bovicola ovis) management practices

Abstract

A deterministic model for predicting the prevalence of lice in sheep flocks was compared with a stochastic model. The stochastic model was based on data from 100 000 simulated farms and encompassed: a range of lice detection levels at shearing, whether or not treatment was applied where no lice were detected, the efficacy of treatments used, and the adequacy of quarantine measures against entering sheep. The deterministic model generated more extreme estimates of the effects of lice management changes, and was judged unsatisfactory for long-term decision making. Predictions based on the stochastic model were consistent with reported survey values, suggesting the stochastic model could be used to predict benefits and costs associated with changes in louse-control methods. According to the stochastic model, the two factors that could provide the greatest reduction in dollar costs were improvement of quarantine by limiting the incidence of lice entering the flock, and improvement of eradication methods. The deterministic model suggested that there would be large benefits from improvement in detection, but the stochastic model only found small dollar benefits with improved detection in the absence of changes such as more effective treatments. Changing the proportion of flocks that were treated when no lice were detected had little effect on the predicted incidence of lousy flocks.