On damage, uncertainty and risk in supervised control : aphids and brown rust in winter wheat as an example

In supervised control of pests and diseases real-time information on pest and disease intensity and cost-benefit analysis of decision alternatives are used to support farmers' decisions on pesticide application. This thesis focuses on improving supervised control in the pathosystem aphids - brown rust - winter wheat. First, effects of the grain aphid, <em>Sitobion avenae</em> , on crop physiology are studied to arrive at simple relations between aphid density and damage. Second, the consequences of uncertainty about the costs of alternative strategies for optimal timing of chemical control of aphids and/or brown rust ( <em>Puccinia</em><em>recondita</em> ) <em>are</em> evaluated.Uptake of assimilates by <em>S. avenae</em> and reduction of the rate of net leaf photosynthesis by honeydew were quantified and introduced into a model which simulates growth and development of winter wheat from flowering till ripeness. Attainable yields from 4000 to 10000 kg ha <sup>-1</SUP>and damage at the higher yield levels were simulated satisfactorily. Honeydew caused more damage than assimilate uptake at crop development stages until milky ripeness. Attainable yield, crop development stage, and timing and intensity of infestation were of major importance in explaining variation in damage. The results of the simulation model were summarized in regression models which predicted damage at given infestation levels as good as the best published empirical model.Sources of uncertainty in a decision model for evaluating costs of chemical control strategies were quantified using field data. Assessment of the contribution of various sources of uncertainty to model output uncertainty showed white noise in the relative growth rates of both aphids and brown rust to be the major cause of uncertainty. Risk-neutral damage thresholds representing pest intensities at which expected costs of control just exceed expected costs of no control, were considerably lower than the deterministic damage thresholds based on average input values. At the risk-neutral damage thresholds only 30% of chemical applications was found to be cost-effective. To support pest control under uncertainty a framework is proposed in which alternative strategies are evaluated in terms of 'profitability' and 'risk', operational measures of the economic objectives 'return on investment' and 'insurance', respectively.