UNCERTAINTY ANALYSIS APPLIED TO SUPERVISED CONTROL OF APHIDS AND BROWN RUST IN WINTER-WHEAT .1. QUANTIFICATION OF UNCERTAINTY IN COST-BENEFIT CALCULATIONS

Information on the uncertainty about predicted costs and benefits of chemical control is an indispensable, but usually ignored, aspect of decision support in crop protection. The consequences of ignoring uncertainty are illustrated using a model for evaluating financial loss associated with different strategies of chemical control of aphids (especially Sitobion avenae) and/or brown rust (Puccinia recondita) in winter wheat in The Netherlands. Crop development, population dynamics and damage are simulated as a function of predicted average daily temperature and a time sequence of decisions on chemical control. The model is initialized with temperature sum and incidences of aphids and brown rust. Uncertainty about model output is computed from uncertainty about model parameters and inputs which are quantified using field data. The probability distribution of financial loss is estimated by stratified sampling from the input distributions followed by Monte Carlo simulation. Compared to no control, chemical control of both aphids and brown rust may reduce the range of possible financial loss. Because the probability distributions of financial loss associated with no chemical control have long right tails, uncertainty leads to damage thresholds well below the deterministic thresholds, even when indifference to risk is assumed. In comparison, the damage thresholds currently used in the decision support system EPIPRE are risk-avoiding for aphids, and approximately risk-neutral for brown rust.